By Andrew Pearce, Alex Kouptsov, Tim Nolan and Jordan Luft

Foreword | Inside AI Transformation Series

This whitepaper is the first of a broader Inside AI Transformation series, which examines how organizations are approaching AI transformation from the inside out. Informed by perspectives from experts at companies participating in BCG’s Spheres AI forum, the series explores the practical organizational choices that determine whether AI moves beyond experimentation and delivers scaled enterprise value. Each paper focuses on a different dimension of internal transformation, including workforce readiness, operating model redesign, adoption discipline, governance, and change management.

Introduction | The Illusion of the Agentic Enterprise

In boardrooms across industries, executives are sketching the same future, a company powered by intelligent agents, where workflows are automated, decisions are augmented, and human effort is reserved for higher-order thinking.

But while the destination is easy to imagine, the path remains elusive.

Many companies have moved quickly from AI experimentation to plans for AI-enabled operating model redesign. Yet inside many organizations, the reality remains uneven. AI tools are deployed but not consistently used. Employees experiment but do not fully adopt. Workflows become fragmented rather than streamlined. What begins as transformation often stalls as a collection of disconnected pilots.

When leaders look for the culprit, workforce readiness often emerges as a central constraint. For many employees, AI adoption requires more than learning a new tool. It asks them to rethink how they work, how they make decisions, and in some cases how they define their professional identity. That challenge is especially acute in knowledge-intensive businesses, where expertise is closely tied to craft. For professionals such as lawyers, journalists, tax specialists, and analysts, AI does not simply change the workflow. It changes the nature of the work.

Companies that scale AI successfully tend to recognize this early. They communicate transparently about how work will change, invest in capability building, redesign workflows rather than simply automate tasks, and treat change management as a core performance discipline.

That is the emerging lesson of the agentic enterprise. The transition is not only technical, but also organizational. Without a deliberate effort to bring employees along, even the most advanced technologies struggle to take hold.

A workforce-led model for scaling AI

In BCG’s experience, companies that scale AI adoption successfully manage four conditions in parallel.

Together, these conditions form a practical model for moving from AI experimentation to scaled impact. They also shift the central question for leaders. The issue is not simply whether an organization has access to advanced AI tools, but whether its workforce is prepared, motivated, and supported to use them in the flow of work.

The following case study shows how these conditions can appear in practice inside a knowledge-intensive enterprise.

Case study: Applying a workforce-led model in a knowledge-intensive enterprise

Thomson Reuters provides one example of how a knowledge-intensive organization has approached this shift. In a business built around specialized professionals, including lawyers, journalists, tax experts, and analysts, the company’s AI transformation required employees to rethink how expertise would be applied as routine tasks became more automated and workflows became increasingly AI-enabled.

Leading with transparency

Rather than minimizing the impact of AI, the leadership communicated early that AI would change roles across the organization. Some roles would evolve. Others would look fundamentally different over time.

At the same time, leadership paired this message with a stated commitment to provide employees with tools, training, and support as roles and workflows evolved. This combination of transparency and investment was intended to help reduce uncertainty and support employee engagement during the transition.

Thomson Reuters reinforced that posture by connecting AI transformation to cultural norms employees already recognized. Instead of positioning AI as a separate change agenda, the company applied existing values to new ways of working. Act fast. Learn fast’ encouraged teams to experiment quickly, learn from failures, and refine approaches based on real-world feedback. ‘Challenge (y)our thinking’, created permission to question assumptions about how expert work should be done and where human judgment should remain central. ‘Stronger together’, reinforced the importance of collaboration across leaders, practitioners, technologists, and change champions as teams navigated new tools and redesigned workflows.

Turning employees into participants

That posture then had to be translated into practice. Many enterprise change programs are designed top-down, with leadership defining a future state and cascading it through the organization. In this case, Thomson Reuters also emphasized employee participation.

Senior leaders shared their own AI experiments, applications, and limitations openly with their teams thereby modeling the experimentation they were asking employees to embrace. In parallel, individual contributors at every level were encouraged to test ideas, build prototypes, and feed their learnings back into the organization. The result was a change effort that combined leader-led and colleague-led elements, which may have contributed to adoption momentum.

Letting workflows shape the tooling

The same logic shaped TR’s approach to tooling. Rather than mandating a fixed set of approved AI tools at the outset, the company allowed teams and individuals to identify what worked for their specific workflows. Exploration was treated as a feature of the transformation path. The decision reinforced the broader signal that employees were trusted to define how AI fit their work, rather than being handed a predetermined answer.

Over time, this approach reduced one of the most significant sources of friction. Employees were not positioned as passive recipients of change. They became active participants in shaping how AI was applied in their work.

Where the model showed results

This workforce-focused approach was associated with reported improvements across several parts of the business. In go-to-market teams, content operations, and the General Counsel’s Office, Thomson Reuters redesigned workflows around AI rather than simply adding tools to existing processes.

The reported results were not uniform across every team or function. They were strongest where employees actively incorporated AI into their daily work, suggesting that adoption intensity mattered as much as tool availability.

Taken together, these examples suggest a consistent pattern. In these examples, value did not appear to come from deploying AI alone. It appeared to emerge where employees adopted the tools, developed confidence in the outputs, and integrated them into how work was done. The distinction between users and non-users was often more significant than the technology itself.

The Hidden Lever: Change Management

Even with strong use cases, Thomson Reuters encountered familiar challenges.

Adoption did not happen automatically. In some cases, usage lagged even where tools were broadly available and actively encouraged. Teams reported tool fatigue as multiple AI solutions were introduced simultaneously. Integration with existing systems added complexity.

What appeared to distinguish higher-performing groups was not access to better technology, but stronger change management.

These groups established clear expectations for usage. They introduced AI champions to support adoption. They provided guidance on when and how to use different tools. They created visibility into usage metrics and reinforced adoption through recognition programs.

In effect, they treated change management as a core operational discipline. In BCG’s experience, this discipline often works as a flywheel: leadership sets expectations, employees experiment in real workflows, champions and embedded experts help translate use cases into better ways of working, adoption data reveals what is working, and the organization uses those insights to refine tools, training, and incentives. Over time, each turn of the flywheel increases confidence, usage, and measurable value.

Thomson Reuters’ Forward Engineering squads provide one example of this flywheel in practice. These internal teams of AI engineers were embedded directly into business functions to redesign workflows alongside the employees who own them.

The model differs from some common patterns in enterprise change. Rather than presenting employees with a designed future state to adopt, Forward Engineers arrive without a predetermined solution. They bring deep technical fluency in AI and the employee brings deep knowledge of the process, its constraints, and the desired outcome. The solution is co-created.

In practice, this means an employee can describe a workflow problem they have lived with for years, without needing to know how AI might solve it, and a Forward Engineer can translate that conversation into something AI-native and human-led. The dynamic changes the experience of change itself. It is no longer something done to the workforce, it is something done with it. For Thomson Reuters, Forward Engineering served as both a change-management mechanism and a way to connect technical capacity directly to employee-identified workflow needs.

The lesson is straightforward but often overlooked. AI transformation does not fail because of poor technology. It fails when organizations underestimate the effort required to change how people work.

5. Redefining the Role of the Employee

As AI becomes embedded in workflows, the role of the employee is evolving. Work that was once defined by execution is increasingly defined by oversight. Employees review and validate outputs generated by AI systems. Judgment becomes more important than repetition.

At the same time, the nature of work shifts toward exception handling. Routine tasks are automated, while human effort is directed toward complex or ambiguous scenarios. There is also a growing expectation of system fluency. Employees must navigate multiple tools, understand how outputs are generated, and move seamlessly between systems.

These shifts represent a meaningful departure from traditional role definitions. They require new skills and, in many cases, a new mindset. Over time, they also begin to reshape how organizations themselves operate, prompting a reconsideration of how roles are defined, how performance is measured, and how employees are supported through change.

For organizations with deeply specialized talent, this transition is particularly significant. Professionals are not only learning new tools. They are redefining how their expertise is applied.

The Road Ahead

The fully agentic enterprise is still taking shape, but the direction is becoming clear. Workflows will continue to automate, outputs will become more consistent, and organizations will unlock new levels of efficiency. At the same time, the role of the employee will grow in importance, not diminish. As AI takes on routine tasks, employees will increasingly focus on oversight, judgment, and orchestrating work across systems.

The Thomson Reuters example illustrates one version of this shift. More broadly, the future is unlikely to be defined by companies without employees, but by companies in which employees operate differently and contribute in more strategic ways.

For enterprise leaders, the implication is increasingly clear. AI transformation is not ultimately defined by the sophistication of the technology, but by how effectively organizations prepare their workforce to adopt and work alongside it. Companies that invest in capability building, align incentives, and treat change management as a core discipline will move faster and realize greater value.

The organizations that succeed in the agentic era will not simply be those with the most advanced AI. They will be those with the workforce best prepared to use it.

Appendix

Methodology / Source

This article was developed by BCG in collaboration with Thomson Reuters. Thomson Reuters provided examples from its internal AI transformation. BCG retained editorial review and responsibility for the broader analysis and implications.

This article was informed by Thomson Reuters. Thomson Reuters examples are used as case illustrations, not as a benchmark or ranking of AI maturity across companies.

About Thomson Reuters

Thomson Reuters is a global provider of information and technology solutions for professionals across legal, tax, audit, accounting, compliance, government, and media.

About BCG

Boston Consulting Group is a global management consulting firm that partners with leaders in business and society to address their most important challenges and capture their greatest opportunities. BCG helps organizations accelerate AI-enabled transformation, redesign operating models, improve performance, and build the capabilities needed to scale innovation and growth.

Contributors

Andrew Pearce: Andrew is Head of Customer Experience and Digital at Thomson Reuters, where his work includes customer experience, digital sales, operational delivery, and AI-enabled transformation

Alex Kouptsov: Alex is a business operations and transformation leader at Thomson Reuters focused on AI-enabled productivity and operating model design

Tim Nolan: Tim is a Managing Director and Senior Partner at Boston Consulting Group. He co-leads BCG’s information services practice, helping companies navigate complex transformations, accelerate AI-enabled growth, redesign operating models, and improve commercial and operational performance

by Jens Uehlecke, Nadja Saller, Ryan Jacobson, Sana Qureshi, Brenda Farrell

In the previous article “Germany: One of Europe’s Largest — and Fastest-Growing — TV Streaming Markets” we established that Germany’s television streaming market is at an inflection point — structurally primed for growth by regulatory change, infrastructure investment, and the accelerating decline of cable and satellite. The market’s competitive landscape has bifurcated between telco-distributed incumbents and agile OTT pure players, and that advertising revenues are increasingly up for grabs as linear budgets migrate toward digital and CTV. But structural opportunity alone does not determine who wins. The more fundamental question — one that market-level analysis cannot answer in isolation — is who, exactly, will migrate, and what it will take to convert them. In this second article of our two-part series, we turn to the consumers at the center of Germany’s streaming transition.

Drawing on BCG proprietary survey research across more than 2,000 Germans, we have identified five distinct consumer segments that define viewing behavior, switching readiness, and purchasing criteria across Germany’s legacy television households. Four of these are addressable live TV streaming segments and we will quantify the revenue opportunity each represents for providers seeking to capture the next wave of growth in Germany. The fifth segment, i.e., the nonlinear streamers watch predominantly on demand and are not the focus of this article.

How Age Matters for Viewership Habits

How Germans watch television varies significantly by age and platform. Daily live television viewing increases consistently with age — from just 6% of 16–24-year-olds watching live TV every day, to 77% of those aged 75+. Yet despite these generational differences in viewing intensity, Live TV Streaming penetration remains substantial across all age groups, hovering above 20% in every cohort as seen in Exhibit 1. This confirms that the Live TV streaming opportunity exists across all demographics.

These differences in how Germans watch Live TV, however, are only part of the picture. Equally important for providers is understanding how viewers engage — specifically, how they begin their viewing session. This divides into three distinct types: channel-led (turning on the TV and browsing channels), content-led (searching for something specific), and platform-led (opening an app and browsing).

As Exhibit 2 demonstrates, younger viewers (ages 16–24) tend to be platform-led (48%), middle-aged viewers (ages 45–64) are more likely to be channel-led, and elderly viewers (75+) are content-led (59%) by a large margin. However, when we combine our results for the online and telephone surveys for 75+ cohort, the viewers are more evenly split across channel-led and content-led but with latter still dominating. What this means is that winning the market requires winning multiple, distinct audiences — each on their own terms.

The four addressable consumer personas and strategic implications

To further deep dive into what distinguishes Live TV Streaming consumers from each other, we have identified four distinct addressable segments that shape Germany’s Live TV Streaming market, each with a different strategic role across conversion, retention, and upsell:

1. Intentional TV Streamers 11% of the market — shares reflect survey panel composition, see footnote 1): They deliberately opt for television streaming as a flexible content platform, streaming from all kinds of devices. They have the highest willingness to pay (€52/month) but majority of them would accept ads for a lower price. Their viewing is selective rather than habitual, and 80% still watch live TV at least several times per week driven by live events. 58% find news channels important, indicating that these consumers value news alongside sports and entertainment. These streamers fall within ages 25–44 across genders.

Strategic Recommendation: Offer premiumization through tiered pricing and multi-device feature sets. High willingness to pay coupled with high ad-tolerance indicates they’ll opportunistically trade down for a discount; hence an ad-supported tier can be a logical fit for this group.

2. Converted Traditionalists (17% of the market): They have migrated to television streaming but remain strongly linear in behavior. 69% watch live TV daily and channel-led viewing dominates, indicating that they use television streaming as a replacement for cable, and have not adopted a new viewing model. The subscription relationship is stable with low subscription aversion and strong upsell headroom (€10/month above current spend). News channels are a core content anchor as 70% of respondents deem them important. Tech quality or reliability (76%) is the dominant purchasing criteria. This segment is more likely to be female and above 45.

Strategic Recommendation: Focus on retention through price-value balance while upselling through tiered packages, given that this segment is largely satisfied with current subscriptions and values a strong channel line-up. Since 75% would accept ads for a lower price, a tiered pricing model is viable but must be positioned carefully to avoid undermining perceived value.

3. Open Legacy Viewers (14% of the market): They still watch television via cable or satellite, but 77% express willingness to switch, and are willing to pay up to €40. They consume a hybrid of streaming and live TV and are channel-led in their viewing habits. Live television watching is driven by habit instead of live events. News channels are relevant but not the most important forms of content. The barriers here are practical-price and setup complexity-rather than attitudinal. 23% are open to additional subscriptions if they are cheaper. These consumers are between ages 16 to 44, and skew male.

Strategic Recommendation: Convert through simple, affordable entry offers with strong channel coverage given their strong inclination to channel-led viewing, making channel breadth the strongest conversion lever. Given their high tolerance for ads, an ad-supported tier could be an attractive conversion lever.

4. Traditionalists 1. (38% of the market): The largest segment of viewers. They are passive, habit-driven, linear-first viewers- 72% watch live TV daily out of routine, and live TV constitutes 53% of total screen time. 74% find news channels important, meaning that news is a non-negotiable content anchor. They have low intrinsic motivation to switch, but they are structurally exposed to change as cable declines. 63% indicate openness to switching when prompted, with tech quality or reliability emerging as the most named purchasing criteria. This segment is usually above age 65.

Strategic Recommendation: Players must position themselves to be the default destination when decline of legacy infrastructure forces migration. A simple, affordable entry offer can unlock this segment but in addition to a low price point, it is critical to also lead with reliability, quality, and ease of setup.

The remaining 18% of the market is characterized by “nonlinear streamers” who do not watch much live TV and instead opt for streaming content on demand. They tend to be younger, and their viewership is platform-driven, rather than by content or channels, and are not anchored to a television set. Lastly, the remaining 2% of households did not map cleanly to any of the five segments and are excluded from the addressable market analysis.

Key purchase criteria: price and flexibility

Overall, price and flexibility, as Exhibit 4 showcases, are the top purchase criteria across all consumer segments, regardless of age, current reception technology, or viewing behavior. This makes value-for-money the non-negotiable baseline for any television streaming proposition.

Beyond price and flexibility, the secondary criteria vary depending on age. Ease of use and setup is critical for Traditionalists (61%) who are the largest market segment, while mobility and multi-device access matter most for younger, more digital segments. Channel breadth is important across all groups on cable or satellite, while entertainment package breadth is a stronger differentiator for both Intentional TV Streamers and Open Legacy Viewers relative to other segments.

Meanwhile, ad tolerance is uniformly high across all segments-between 75% and 79%-validating a tiered pricing architecture that includes an ad-supported entry tier across the full market, not just price-sensitive audiences.

Value for money is the universal churn driver

To win in this market, providers must retain what they already have while also converting legacy consumers. For Intentional TV Streamers and Converted Traditionalists, the two largest existing live TV Streaming consumer segments, retention depends on sustaining both usability and value for money — weakness in either dimension materially increases churn risk. Meanwhile, Traditionalists require an external trigger due to satisfaction/inertia; Open Legacy Viewers require a practical trigger (price/simplicity offer), not an attitudinal one.

Overall, across all consumer segments, value for money emerges as the single largest churn driver as seen in Exhibit 5. No group, regardless of loyalty or engagement, will tolerate poor value indefinitely. Low usage typically ranks second, signaling that customers who never fully engage with the product are the first to leave once price pressure mounts or a trial period ends.

Near-term growth pool: 14.1 million switch-ready legacy households

Germany’s legacy television base contains a core addressable migration pool of 14.1 million households that have already indicated readiness to switch. This is made up of two segments: 4.4 million Open Legacy Viewers — 77% of that segment’s 5.7 million households — and 9.7 million Traditionalists, or 63% of that segment’s 15.4 million households. At current willingness-to-pay levels, the full switch-ready pool represents a revenue ceiling of approximately €4.4 billion annually. This is incremental to today’s €2.6 billion subscription market — not a restatement of it — and is consistent with the €6.2 billion base-case trajectory modeled in the first article of this series.

However, applying the telephone survey’s more conservative switching rate to the Traditionalist segment — 20% actively open versus 63% in the online panel — yields a risk-adjusted near-term pool of approximately 7.5 million households and a realistic revenue opportunity of roughly €2.3 billion. The gap between the two figures is not a reason to discount the opportunity; it defines the activation challenge: the €2.1 billion difference sits in households that are structurally exposed to switching but not yet actively seeking it, reachable through the right trigger rather than the right offer.

Conversion readiness depends on what kind of trigger ultimately moves a household to switch. Active switchers — primarily Open Legacy Viewers and younger Traditionalists — respond to price and product signals, meaning providers can win them through targeted offers. Forced switchers — primarily older Traditionalists — will only move once cable infrastructure leaves them no other option. Because these two populations are motivated by fundamentally different forces, they require distinct acquisition strategies and onboarding experiences to convert successfully.

Conclusion

In our first article, we established that Germany’s television streaming market is at an inflection point: cable and satellite are in structural decline, regulatory tailwinds have removed the last structural barrier to switching, and both growth scenarios we modeled point to substantial subscriber and revenue growth through 2035. We also showed that the competitive landscape is bifurcated — between telco-distributed incumbents that lead on scale and OTT pure players that lead on growth momentum — while the advertising market is increasingly shifting to digital.

However, structural opportunities do not automatically translate into commercial success. This article explores why by examining the preferences, behaviors, and needs of the German consumer. We identified four distinct consumer segments- Intentional TV Streamers, Converted Traditionalists, Open Legacy Viewers, and Traditionalists — each with different switching readiness, different content anchors, different purchasing criteria, and different churn drivers. No single proposition captures all four. Providers that treat the migration pool as a homogeneous addressable market will underperform those that match offer design to segment-specific barriers.

The near-term priority is clear: the 14.1 million switch-ready legacy households represent the structural ceiling of the medium-term migration opportunity, with a revenue potential of approximately €4.4 billion — though the telephone survey correction puts the actively reachable near-term pool at closer to 7.5 million and €2.3 billion. The gap is not a reason for caution; it defines the work. Converting these households requires precision: price and flexibility are the universal baseline, but the winning proposition varies by segment — simplicity and reliability for older legacy viewers, channel breadth for cable migrants, multi-device access for younger digital audiences. Providers that operationalize this segmentation in product design, pricing architecture, and go-to-market execution during the current migration window are best positioned to capture Germany’s next wave of streaming growth.

Germany is one of the largest market opportunities in Europe for television streaming. While some European markets have matured and others are in the early stages, Germany has been rapidly growing and is poised for further acceleration. Europe’s largest total TV market by household count, Germany is the fastest-growing major European television streaming market at 12% CAGR (2020–2025), outpacing the top five markets by 10 percentage points. In line with global trends, the market is undergoing a major structural transformation from legacy television to television streaming. Television streaming — the delivery of live and on-demand television over an internet connection — has reached 18% of German households. It is the only distribution technology that is growing, while cable and satellite have been in long-term decline across Europe.

Importantly, the primary barrier to adoption is not consumer awareness but inertia, reframing the strategic challenge from one of “market creation” to “market activation”. As households increasingly migrate away from legacy platforms, the key question is no longer whether the transition will occur, but which players will capture the migrating customers and associated value pools. This creates both a significant opportunity and a narrowing window for action: market participants must not only develop a deep understanding of the market dynamics but also move quickly to capitalize on them. Whether an established telco with an in-house TV streaming service or an OTT player with a strong B2C or B2B business, the opportunity is substantial. The winners will be those that act early to secure customers, partnerships, and strategic positioning as Germany’s television ecosystem continues its transition toward streaming. Delayed action could limit value capture as the market becomes increasingly crowded and competitive.

Drawing on BCG proprietary research (expert interviews, online and telephone surveys) and third-party data, we identified three strategic priorities that market participants must address to position themselves for success and capture this emerging opportunity:

1. Understand Bifurcated Competition: The television streaming competitive landscape is bifurcated between telco-distributed players and OTT pure players. Understanding how they differ on product, monetization, and positioning is essential for any market participant.

2. Adapt Advertising Strategies: The €4 billion German television advertising market is migrating from linear to digital, which offers both broadcast reach and digital targetability. But advertisers compete in an increasingly crowded digital landscape, so advertising revenue must be won through targeting capability and reach, not just audience size.

3. Target Consumer Personas: Consumer behaviors and purchasing decisions are not uniform. Four distinct personas — with different viewing habits, switching readiness, and purchasing criteria — define who the market can reach and how. Converting them requires matching the right offer — based on price, simplicity, channel breadth, and reliability — to the specific barriers each persona faces.

This article, the first in a two-part series, focuses on the market dynamics shaping Germany’s television streaming landscape. We examine the bifurcated competitive structure between telco-distributed and pure-play OTT providers, as well as the evolution of advertising as audiences shift from linear TV to digital streaming environments. In a follow-up article, we turn to the consumers of the market, exploring the four consumer personas emerging from our research that define viewing behavior, purchasing decisions, and growth opportunities in Germany’s streaming ecosystem.

Defining the Market Opportunity in Germany: Europe’s fastest-growing TV streaming market

Before exploring the strategic imperatives for market participants, it is important to first understand Germany’s television streaming landscape and the forces driving the opportunity. Therefore, in this section, we will outline the market’s size, growth trajectory, and underlying dynamics to provide context for what it will take for market participants to succeed.

Since 2015, television streaming has more than doubled its share of German TV households, increasing from 8% to 18%, while both cable and satellite have declined, currently accounting for 36% and 43% of the market respectively. There are four instrumental forces driving this shift:

1. Rising digital adoption
2. Rollout of fiber and 5G infrastructure
3. Television streaming’s superior position on price and flexibility, and
4. Successive regulatory changes that have dismantled the structural advantages of cable. A key structural catalyst has been the abolition of the Nebenkostenprivileg (NKP) — a law that allowed landlords to bundle cable fees into utility costs — in July 2024, forcing cable households to actively choose a TV provider for the first time.

Zooming out to examine broader European trends provides insight into where Germany is heading. Streaming adoption across Europe remains highly fragmented. Terrestrial heavy markets such as Italy and Poland continue to exhibit single-digit penetration rates, while markets with strong incumbent telcos that bundle TV with broadband — including France, Portugal, and Switzerland — have reached penetration levels of 56–69%.

At 18% penetration, Germany’s television streaming market remains well below these more mature European markets. However, the gap is narrowing, suggesting that Germany is entering the acceleration phase of its own S-curve. As shown in Exhibit 1, Germany recorded a 12% CAGR between 2020 and 2025, making it the fastest-growing major television streaming market in Europe and outperforming the average growth rate of the five most advanced European TV streaming markets by approximately 10 percentage points.

Based on the market dynamics outlined above, we have come up with two likely scenarios over the next decade with regards to the evolution of Germany’s television streaming market. A “base scenario” and a “momentum scenario.” While the pace of adoption may differ across these scenarios, both point to significant opportunity for market participants that position themselves effectively.

The “base scenario” forecasts strong market growth through subscriber expansion alone. In this scenario, the current structural trends continue, with television streaming reaching 42% household share by 2035, while both cable and satellite decline gradually. Subscription revenue grows from €2.6 billion (2025) to €6.2 billion (2035), representing a 9% CAGR.

The “momentum scenario” envisions a significant expansion of the streaming market facilitated in part by accelerated decline of legacy technologies. The cable infrastructure is phased out post-2030 (no new additional customers), and satellite declines even more rapidly due to broadband expansion and reduced provider support. Television streaming captures most of the switches and reaches 58% of households by 2035, which is 6.5 million additional households compared to the base scenario. Subscription revenue grows at ~12% through 2035, reaching €8.5 billion.

As shown in Exhibit 2, both scenarios exhibit an S-curve adoption pattern: strong near-term growth driven by the migration of legacy television households, followed by a gradual slowdown as the remaining holdout segments become progressively harder to reach. Regardless of which scenario emerges, both present a substantial opportunity for market participants to capture migrating households and associated value pools.

I. Understand the Bifurcated Competition: Telco Giants and OTT Pure Players

The European TV streaming markets are led by telco-distributed live-streaming players, while the OTT pure players remain subscale in absolute terms.

Telco-distributed players own the largest subscriber bases by bundling TV with broadband contracts. These platforms monetize almost exclusively through subscriptions and benefit from existing customer relationships and bundle lock-in. In Germany, these players include MagentaTV (Deutsche Telekom), GigaTV (Vodafone), 1&1 TV, o2 TV, and PYUR TV.

Meanwhile, OTT pure players that deliver content over the open internet without requiring a telco relationship, typically combine subscription and advertising revenue. These players tend to offer broader channel lineups at competitive entry price points and have consistently posted significantly higher growth rates than their telco counterparts. In Germany, these players include waipu.tv, Zattoo, Sky Stream, and Joyn (hybrid model with own content). In fact, waipu.tv stands out as Europe’s largest OTT pure player, leading by subscriber count and ahead of peers in France (Molotov), Spain (Agile TV), Switzerland (Zattoo), the Netherlands (NLZiet), and Poland (Player.pl, Pilot WP). That makes Germany a reference market for OTT-led television streaming development in Europe.

Overall, Germany’s market structure suggests that incumbency alone is not enough to win. While telcos currently lead in scale, the strong growth of OTT pure players demonstrates that opportunity remains up for grabs. As pricing converges — currently at around €5–7 per month across both OTT pure players and several telco-distributed offerings — and streaming adoption accelerates, competitive advantage will increasingly shift from price and distribution ownership to product differentiation, content, partnerships, and customer experience. Hence, the next phase of growth remains contestable for both telco and OTT players.

II. The Advertising Opportunity: From Linear to Digital

While subscription revenues remain the primary monetization lever for television streaming platforms, advertising is emerging as an increasingly important source of incremental growth. Germany’s television advertising market has held broadly stable at approximately €4 billion, but there nevertheless has been a decisive compositional shift. As Exhibit 3 shows, linear FreeTV remains the dominant format at 80% market share, but it is declining at roughly -4% CAGR, while digital video formats are growing at ~25% CAGR from a much smaller base.

Streaming platforms have an opportunity to monetize their growing viewer bases through premium video inventory and targeted advertising solutions. They are already exploring Dynamic Ad Substitution (DAS) in live TV, Dynamic Ad Insertion (DAI) in on-demand streams, and display/EPG formats. This adds high-margin incremental income on top of the subscription revenue that continues to be a major revenue source for streaming platforms. This hybrid model structurally strengthens unit economics as subscriber scale grows.

Meanwhile, Connected TV (CTV) — video delivered over the internet to smart TVs and connected devices — is expanding television into a fully addressable advertising channel. CTV ad spend in Germany is projected to grow more than eight-fold from 2020 levels to USD 2.7 billion (~EUR 2.36 bn) by 2030, mirroring the trajectories in the United States.

Germany’s linear-to-CTV advertising crossover point is projected to take place around 2032 — a four-year lag behind the United States, where CTV is expected to overtake linear TV by 2028. Platforms that build scaled, addressable inventory now are best positioned to capture reallocating budgets as the crossover point from linear to CTV approaches.

Unlike linear TV, where the advertising market was largely contained within a defined set of broadcasters, digital video advertising is contested by a much wider range of players: social video platforms (YouTube, TikTok, Instagram), SVOD services with ad tiers (Amazon Prime Video, Netflix), and dedicated CTV operating system players (Roku, Amazon Fire TV). These players bring structural advantages in data and targeting. Amazon, for example, leverages first-party retail purchase data to enable closed-loop attribution from ad exposure to purchase — a capability that traditional TV advertising cannot replicate. This creates both a competitive threat and a benchmark for the targeting standards the market is moving toward.

The implication is clear: while television streaming platforms will face intense competition for advertising budgets from hyperscalers and digital-native players with superior data assets, the opportunity remains substantial. By building scaled, addressable advertising inventory and leveraging their unique combination of premium television content, broadcast-scale reach, and digital targeting capabilities, streaming platforms can secure a meaningful share of the growing digital and thereby CTV advertising market. For many players, advertising will increasingly serve as a critical complement to subscription revenues, strengthening overall platform economics and creating an additional avenue for value creation.

Conclusion

Germany’s television market is undergoing a structural transformation, positioning it as the largest market opportunity in Europe within the near-term. The distribution shift is not simply a consequence of regulatory changes; it is the product of a decade of converging forces — consumer habit change, infrastructure investment, product improvement, and regulatory liberalization — that are unlikely to reverse.

The competitive landscape will be shaped by the interplay between telco-distributed scale and OTT agility. Telco players hold the largest bases but are constrained by bundle economics and limited advertising monetization. OTT pure players are growing faster, monetizing more broadly, and as the European benchmark confirms, Germany’s own OTT market is more advanced than European peers.

The advertising opportunity amplifies the subscriber story. As linear budgets migrate to digital and CTV inventory scales alongside subscriber bases, the market becomes attractive on both dimensions simultaneously. But television streaming platforms now compete in a broader digital advertising ecosystem alongside social platforms, SVOD players with ad tiers, and CTV OS operators — meaning that advertising revenue is won through targeting capability and reach, not just audience size.

The next phase of competition will be determined not only by how providers compete, but also by whom they target. Our consumer research reveals that Germany’s legacy television households are far from homogeneous, exhibiting distinct viewing habits, switching readiness, and decision-making criteria. While some consumers are actively evaluating alternatives and motivated by flexibility and price, others remain deeply attached to traditional viewing habits and place greater emphasis on reliability, technical quality, and familiar channel navigation. These differences affect both switching propensity and willingness to pay.

As Germany enters the acceleration phase of its streaming adoption curve, providers will need to move beyond one-size-fits-all propositions. Success will depend on identifying which households are most ready to migrate, understanding the barriers holding back each segment, and tailoring offers accordingly.

In the second article of this series, we reveal the four consumer personas emerging from our research — from digitally engaged switchers to more traditional television loyalists — and quantify the revenue opportunity, conversion potential, and winning proposition for each. For providers seeking to capture Germany’s next wave of streaming growth, understanding these personas may prove as important as understanding the competitive and advertising dynamics discussed here.

[1] Deutsche Glasfaser, EWE, NetCologne, M-net, westconnect, NetCom BW, TNG, KEVAG Telekom, WOBCOM, STIEGELER, etc.

Germany: One of Europe’s Largest — and Fastest-Growing — TV Streaming Markets was originally published in BCGonTech on Medium, where people are continuing the conversation by highlighting and responding to this story.

By Stephen Edison, Helen Kondos, Brian O’Malley, and Glenn Fiedler

Most corporate legal departments are not yet integrating AI into their core operations. According to the 2026 BCG Legal Industry Survey of corporate legal departments, 91% are piloting AI tools and 93% use general-purpose AI assistants, yet the vast majority have not begun delivering value at scale or redesigning how work actually gets done. The gap is not technological. The tools are there. What is missing is the organizational, commercial, and operational change required to make them matter.

Three questions can help General Counsel diagnose where they stand, and what it will take to move from adoption to impact.

1. Are My People Both Willing and Able to Use AI?

The binding constraint is not tool availability; it is tool adoption. Adoption requires people who are both willing and able. These are distinct problems. A lawyer may be open to AI but unable to evaluate whether its output is reliable. Another may have the aptitude but resist adoption because no one around them is using it, or because they see it as a threat. In the survey, 40% of corporate legal respondents cite trust in AI accuracy as a top-three barrier, and 49% point to workflow integration. The implication is that building genuine familiarity with AI tools is itself a lever: across the full survey, respondents who are regular AI users report significantly lower trust concerns (38% vs. 67%) and more than double the rate of value at scale (34% vs. 15%).

Departments closing the gap deploy a reinforcing set of interventions:

• Structured learning that teaches critical evaluation of AI outputs: not just prompting, but assessing completeness, spotting missed clauses, and iterating to useful results.
• Low-risk experimentation with guided tasks and feedback loops. One bad experience with a tool can lose a user permanently; structured practice builds both competence and confidence.
• Power user mentorship. Resistance erodes when peers demonstrate visible results. Identify early adopters and give them a formal role in bringing others along.
• Visible wins through internal case studies. Concrete examples persuade; abstract promises do not.
• Usage monitoring to identify where adoption stalls and why, tracking not just access, but how tools are used and where value materializes.
• Aligned incentives. If AI fluency is not part of performance evaluations, it remains a side project.

The hard question: Does your team know how to evaluate an AI output: not just whether to accept it, but why it is good or bad and how to improve it? If not, you have tool deployment, not a strategy.

2. Does Outside Counsel Have an Incentive to Use AI, and Pass the Benefit to Me?

A significant share of legal spend flows to outside counsel. Under a billable hour model, firms have limited incentive to use AI; every hour saved is revenue lost. In the survey, 81% of all respondents expect AI to reshape the law firm business model within five years, and among law firms with more than 200 attorneys, 57% report that clients are already asking about AI’s impact on fees. Yet most GCs have not translated that expectation into commercial terms.

Several levers can change the equation:

• Move commoditized work to fixed fees. For routine matters such as standard filings, NDA review, and due diligence, fixed-fee arrangements give firms a direct incentive to use AI: efficiency gains improve their margin rather than eroding their revenue. This is the single most powerful lever for driving outside counsel AI adoption.
• Probe for real investment. When evaluating outside counsel, go beyond talking points. Which workflows have firms actually redesigned? What productivity gains have they measured? The gap between firms that are transforming and firms that are marketing is widening.
• Explore blended rates that reflect AI-assisted delivery. Where junior associate work is materially augmented by AI, the rate for that component should change. Pricing should reflect how work actually gets done.
• Experiment with alternative legal service providers. ALSPs are gaining traction for standardized work, and some are delivering AI-enabled efficiency that traditional firms have not yet matched. For commoditized or high-volume matters, testing ALSP alternatives can reveal meaningful cost and speed advantages.
• Recapture institutional knowledge. Outside counsel holds years of accumulated knowledge about your risk appetite, standard positions, and regulatory history. For departments building internal knowledge systems, recovering that data is a prerequisite, not an afterthought. This starts with clear contractual terms: engagement letters and outside counsel guidelines should explicitly require that all work product, data, and matter-related materials are the property of the client and must be delivered in structured, usable formats upon request.

The hard question: Have you given outside counsel a commercial reason to use AI? If they are still billing you hourly for commoditized work, the answer is no.

3. Have I Evolved How Legal Engages with the Business?

AI does not just accelerate legal work. It changes how legal should interact with the rest of the organization. The highest-performing departments are not simply doing the same work faster; they are rethinking the operating model across three dimensions. And the survey shows where the opportunity is greatest: 74% of corporate legal respondents expect AI to most impact contract drafting and negotiations, 72% point to contract lifecycle management, and 63% cite corporate governance and compliance.

Make legal requirements self-serve.

Business teams either escalate everything to legal, creating bottlenecks, or engage too late, creating risk. AI-powered triage tools and guided intake workflows can help stakeholders assess their own situations against known legal thresholds, so legal gets engaged at the right time, on the right issues, with the right context.

Make institutional knowledge queryable.

Every department has a deep reservoir of prior opinions, contract precedents, regulatory analyses, and negotiated outcomes. The problem is that this knowledge is fragmented across systems, teams, and outside counsel. In the survey, 47% of corporate legal departments cite data readiness as a top barrier to AI adoption, compared to just 30% of law firms. Two things must happen. First, consolidate it: ensure critical knowledge is housed in-house, across all teams and recovered from outside counsel partners. Second, make it queryable. AI-powered search lets anyone in the department investigate what the organization already knows before starting from scratch. Departments that do this stop repeating work and start compounding their institutional advantage.

Automate regulatory impact assessment.

The regulatory environment grows more complex every year. In 2025 alone, 38 states adopted roughly 100 laws regulating AI in some form, according to the National Conference of State Legislatures¹, and over 1,200 AI-related bills were introduced across all 50 states². That is just one domain. Layer on data privacy, employment, financial regulation, ESG, and sector-specific compliance, and the volume of regulatory change that a legal department must absorb is staggering. The challenge is not just awareness; it is speed and accuracy. When a new regulation is passed, how quickly can the department identify the parts of the business it impacts, and how effectively can it assess the implications? AI-powered regulatory monitoring and impact analysis tools can compress what traditionally takes weeks into days: scanning new legislation against the company’s operations, flagging affected business units, and surfacing relevant internal precedents. The value is twofold. On the efficiency side, automated analysis frees legal teams from the manual burden of reading, categorizing, and routing regulatory updates, allowing them to focus on judgment-intensive interpretation and response. On the risk side, speed matters because every week of delay in identifying a regulatory impact is a week of unmanaged exposure. Departments that can assess new regulations quickly and comprehensively reduce the window of non-compliance risk and position the business to adapt ahead of enforcement timelines.

The hard question: Is your legal department still a reactive service function, or has it evolved into a proactive capability the business can access on its own terms?

The Bottom Line

The departments extracting real value from AI did not get there by choosing better tools. They invested in making their people capable, restructured outside counsel relationships to align incentives, and evolved how legal engages with the business. The survey is unambiguous: organizations that have moved beyond pilots into genuine workflow redesign are far more likely to report value at scale. These are organizational and commercial decisions, not technology decisions. The tools will keep improving. The question is whether your department is building the conditions to use them transformationally. The GCs that act fastest will see the benefits compound the soonest: cost savings that fund further AI investment, risk mitigation that builds organizational trust in the tools, operational improvements that free capacity for higher-value work, and a modernized department that attracts AI-native talent.

Source: 2026 BCG Legal Industry Survey

¹ National Conference of State Legislatures, Artificial Intelligence 2025 Legislation; see also Built In, “12 Tech Laws Taking Effect in 2026 You Need to Know,” https://builtin.com/articles/tech-laws-2026

² NCSL data as reported in “Comprehensive List of State AI Laws,” https://stackcybersecurity.com/posts/ai-state-laws

From Pilots to Impact: Three Questions Every General Counsel Should Be Asking About AI was originally published in BCGonTech on Medium, where people are continuing the conversation by highlighting and responding to this story.

As the gaming industry focuses on AI-enabled development, cloud distribution, and evolving monetization models, a deeper structural shift is unfolding. Our gaming survey, including parents’ input on kid’s playing habits, shows that children are entering gaming earlier, engaging more socially, and forming economic expectations years before adolescence.

This is not incremental change. It is reshaping where value accrues across the ecosystem.

Key Takeaways

• Gaming now begins at age 5, which is down from 7 the last time we surveyed two years ago.
• Participation among 10- to 13-year-olds is nearly universal, with many playing primarily to interact with friends.
• Mobile is the primary entry point, but households still plan to invest in dedicated hardware.
• Under-18 players create more digital content than adults, normalizing participation over passive play.
• Younger cohorts over-index on live service and access-based formats.
• Parental trust is not a side issue. It is a growth constraint.

The Starting Line Has Moved

The average starting age for gaming has fallen to five. Thirty years ago, it was thirteen.

This shift expands the formative window during which platform habits, social norms, and monetization expectations are established. Gaming identity now forms in early primary school, before brand preferences fully solidify.

Participation among 10- to 13-year-olds is close to universal. For many, games function less as entertainment products and more as social infrastructure.

First exposure typically occurs on mobile devices. Low friction drives experimentation. But mobile entry does not eliminate platform identity. Many parents plan to purchase a dedicated gaming device within a year, often marking the household’s first hardware investment.

Mobile captures initial engagement. Immersive platforms anchor long term commitment.

Participation Is the Default. Trust Is the Gate.

Under 18 players spend more time creating digital content each week than adults. Building, modifying, and sharing experiences are integrated into gameplay. Creation is not a feature. It is baseline behavior.

Children often encounter brands physically, guided by parents, before transitioning into digital ecosystems. It is this second moment, digital immersion, where long term loyalty forms.

Yet participation runs directly into parental constraints. Parents value creativity and learning, but remain concerned about monitoring, unknown interactions, and time spent.

Platforms that solve for safety at scale expand the total addressable market. Those that do not face structural limits on growth.

Early Norms Are Rewriting the Economics

Motivations evolve generationally.

• Today’s retirees grew up with consoles.
• Millennials grew up with online multiplayer and digital monetization.
• The youngest generation is growing up inside creator economies. For them, UGC environments are not novel, but foundational.

The first game most children under 13 now play is a social platform. Globally that has been Minecraft. In the United States it has recently shifted to Roblox. The first emotional attachment is increasingly to an ecosystem, not a franchise.

This distinction matters: Franchises monetize products. Ecosystems monetize participation.

Younger players over index on social interaction and creative exploration, while older players skew toward achievement and progression. Younger cohorts also demonstrate stronger preference for live service formats and subscription-based access. Preferences that are not cyclical, but conditioned early.

A generation raised in persistent digital environments develops different expectations around ownership, continuity, and value. Access feels natural. Iteration feels expected. Static products feel incomplete.

Monetization shifts are therefore downstream of childhood experience design.

The implications are material:

• Lifetime value curves extend earlier.
• Platform operators capture disproportionate share relative to single IP publishers.
• UGC enabled ecosystems compound engagement.
• Switching costs attach to networks, not characters.

The economic center of gravity shifts from product launches to persistent environments.

A Structural Imperative for Industry Leaders

Children are entering gaming earlier than any cohort before them. They treat games as social infrastructure, create as a default behavior, and internalize access-based monetization before adolescence.

The next decade’s winners will not simply ship better titles. They will design ecosystems that:

• Earn parental trust early
• Enable safe creation at scale
• Transition users from entry level access to immersive ecosystems
• Align monetization with expectations formed at age seven

The future of gaming is not only being shaped by AI or distribution models. It is being shaped in living rooms, years before most players reach their teenage years. The companies that win in 2035 may already be winning with today’s second graders.

Generational Shifts: How the Youngest Generation Is Redefining the Future of Gaming was originally published in BCGonTech on Medium, where people are continuing the conversation by highlighting and responding to this story.

Moving from experimentation to production

As outlined in The Future is Onchain¹, tokenization of real-world assets (RWAs) is a long-term structural shift now underway, as settlement, custody, fund administration, and other activities migrate onto programmable rails. Excluding stablecoins, the total value of distributed assets onchain (from tokenized credit to public equities) has grown from <$100M in February 2021 to $26B in February 2026² — over 250x growth in just 5 years and expanding across multiple asset classes.

Tokenization presents obvious benefits, such as the ability to trade 24/7/365, fractional ownership of assets, global access, and programmability via smart contracts. For financial institutions, blockchain technologies can enable (i) efficiencies in servicing traditional products, (ii) new offers around previously illiquid assets, and (iii) access to a growing base of onchain investor segments.

Institutions are no longer asking whether to build onchain, but rather:

“On which blockchain(s) will institutional finance run?”

Financial institutions are increasingly moving from sandbox experiments to production deployments on public Layer-1 chains³ — tokenized money market funds (TMMFs) such as BlackRock’s BUIDL and Franklin Templeton’s BENJI now hold over $3B in combined total asset value⁴. Blockchain selection has become a critical decision for Digital Assets leaders with real security, operational, and reputational implications. As one executive noted: “Once assets are live onchain, this is no longer a pilot. It is balance sheet exposure.”

Through a series of interviews with 20+ Digital Asset leaders across global banks, asset managers, and custodians (representing over $1T in combined market cap and over $30T in combined AUM), we have distilled the primary decision factors and considerations that underpin the blockchain network selection process.

The institutional decision framework

Based on discussions with Digital Assets leaders, we determined the key evaluation criteria for blockchain selection and their relative rankings, based on their unique onchain use cases. Nine criteria emerged, which fell into four overarching themes. The rankings of these decision criteria in EXHIBIT 1 reflect the relative influence of each criterion, based on interview input (Note: rankings are directional).

1. Risk, control, and compliance: ensuring adequate levels of security, reliability, governance, and privacy such that an institution can operate onchain in a compliant, safe way
2. Technical and economic viability: assessing whether a blockchain is sufficiently performant and cost-effective for a respective tokenization use case
3. Ecosystem traction: determining the health and attractiveness of onchain activity, with participation from developers, institutions, and DeFi
4. Soft factors and influence: weighing the impact of a blockchain’s foundation, culture, and general reputation associated with onchain activity

EXHIBIT 1

I. Risk, control, and compliance

Every institution we spoke with begins its assessment here: Can this network operate within the bounds of our risk, regulatory, and fiduciary constraints? The very features that define blockchains — such as immutability, public transactions, irreversibility — can introduce unacceptable risk when client funds are involved.

Security & reliability

Institutions perform security and reliability due diligence on blockchains with a few core questions in mind: How susceptible is the network to attack or settlement disruption? How does it perform under stress? And how transparent are its incident response processes?

Security
Institutional concerns surrounding security are not unfounded. For example, in 2022, the Ronin Network, a sidechain, had over $600 million siphoned by North Korean state-sponsored hackers, with the breach undiscovered for six days.

Decentralization is widely cited as a proxy for security — the more distributed a network’s validators and stake, the harder it is for any bad actors to disrupt consensus. On this factor, Ethereum is formidable: ~1 million validators and over $70B in staked ETH. But raw scale can be misleading, with the Nakamoto Coefficient (NC)⁵ telling a more nuanced story.

Depending on methodology⁶, Ethereum’s NC can range from 2–5 because of a concentration in staking services. This makes the network theoretically susceptible to a collusion attack between a small number of parties, even if such a scenario remains highly unlikely in practice. Some Layer-2 chains are sequencers operated by a single entity, giving them an NC of 1 — however, the tradeoff for more concentration is that these chains can be managed by reputable, established companies (e.g., Coinbase, as a member of the S&P500, operating Base). Therefore, decentralization must be assessed holistically on multiple dimensions.

As part of their due diligence, many institutions are now conducting deeper security assessments through third parties and audit firms. For example, Halborn is a blockchain-focused security firm that provides services such as smart contract assessments (e.g., code integrity), Layer-1 blockchain assessments, and red team exercises (e.g., simulating real-world attacks).

Reliability
Institutions are also gauging continuous uptime as a key measure of reliability. If a network goes down, client funds and assets onchain are effectively immobilized and inaccessible. While some conceded an occasional brief uptime hiccup (<10 min) would be far from disastrous, the large majority noted network reliability risks as a dealbreaker.

When considering Layer-1 chains, Ethereum’s relative longevity and perception of being “battle-tested” with no downtime over its 10-year history were cited as sources of comfort. For other Layer-1 chains such as Solana, historical outages remain part of diligence conversations, though confidence in the network has increased significantly in recent years through infrastructure upgrades and a demonstrated history of stabilization and resilience under stress.

For institutions, no single metric captures security and reliability factors. Validator counts, NC, total economic stake, and uptime each reveal different dimensions of risk — and the relative importance of each depends on the institution’s specific threat model and use case.

Governance

While governance is not weighted as heavily as security, it remains a critical consideration for institutions evaluating long-term viability. Institutions examine who ultimately decides protocol upgrades and parameter changes. Frequent questions institutions are evaluating are:

• Is the roadmap transparent? Is there a credible trajectory of innovation
• Are incentives aligned among validators, token holders, and core contributors?
• Could governance drift expose institutions to regulatory or reputational risk?

Importantly, institutions must weigh the tradeoffs associated with two distinct governance models. Decentralized governance designs determine upgrades and rule changes through distributed consensus mechanisms — this optimizes for network neutrality and censorship resistance but can introduce uncertainty around speed and stability of decision-making and coordination.

In contrast, consortium-based networks may determine governance and design choices among a defined group of stakeholders or partners. For example, the Canton Foundation brings together major financial institutions to operate a coordinated governance framework for the GlobalSynchronizer, the technical backbone of the Canton Network. This model may offer clearer accountability, more predictable upgrade paths, and stronger alignment with enterprise risk frameworks — but may do so at the expense of perceived neutrality or decentralization.

For institutions accustomed to clear accountability structures, the core question is straightforward: if something goes wrong, who is responsible and who can fix it? This drives Digital Assets teams toward governance models that offer operational predictability and manageable change risks.

Privacy & Permissioning

Privacy
Privacy is one of the most critical gating factors for institutions considering public chains — visibility of transaction flows creates both commercial and regulatory risk. When operating on public chains, privacy requirements vary by use case, institution type, and jurisdiction, making it difficult for Digital Assets teams to define with absolute certainty which privacy features:

1. They need to protect themselves and/or their clients when enabling their use case
2. Different blockchains provide natively or via extensions
3. Satisfy regulatory requirements

A compilation of cited privacy features from interviewees are included in EXHIBIT 2 (non-exhaustive):

EXHIBIT 2

Public chains themselves provide or are actively developing privacy-enhancing technologies (PETs) to fulfill institutional needs. As one public chain example, Solana has launched:

• Solana Permissioned Environments (SPEs) for permissioned instances of Solana Virtual Machine
• Confidential Balances for masking wallet balances and transfer amounts
• Contra for private payment execution channels on top of mainnet
• Token Extensions to embed privacy features such as asset whitelisting directly onto tokens
• SSTS (Solana Security Token Standard) to issue regulated and configurable tokens

These types of offerings can stack and layer onto one another and even exist on layers beyond the blockchain (e.g., in wallets), but there is still uncertainty — among institutions, regulators, or the chains themselves — on whether these or comparable solutions across networks are completely aligned to satisfy regulatory requirements.

On the other end of the spectrum, natively permissioned networks like Canton have privacy as the default — contrary to the innately “visible” nature of public chains. Canton has established a strong market-fit for institutions looking to prevent risks like identity leakage, frontrunning, and the exposure of competitive intelligence for use cases like repos (repurchase agreements). However, there are tradeoffs to consider regarding broader access and distribution potential to the ecosystem of users and developers when compared to natively-public chains.

Permissioning
Despite public chains’ privacy feature offerings, permissioned or hybrid architectures retain appeal given some institutions’ desire to pilot or sandbox ahead of a public deployment. Many banks described a phased approach — experimenting in permissioned environments like Hyperledger Besu before moving to public settlement layers — while others favor hybrid designs where execution stays on controlled, permissioned infrastructure but settlement occurs on public chains, giving tokenized assets access to broader onchain liquidity.

The dividing line often comes down to use case. For some payments and RWA use cases, public networks provide ecosystem access and liquidity but may lack some privacy features. For repurchase agreements, bilateral treasury movements, or interbank settlement, private and permissioned environments remain attractive where complete identity and position masking are essential.

II. Technical and economic viability

Performance

Early blockchain infrastructure was not viable for institutional use cases due to slower throughputs — consider Bitcoin’s ~7 maximum theoretical transactions per second (TPS)⁷. However, for many onchain institutional use cases today (e.g., money market funds, corporate bonds, tokenized debt issuances), the majority of major chains have sufficient throughput (1k+), often negating TPS as a differentiating factor.

What may be more important is resilience under stress — for example, January 31, 2026, saw over $2.5B in crypto liquidations in 24 hours due to market volatility, testing the capabilities of public networks. As shown in EXHIBIT 3, Solana’s TPS nearly tripled during peak volatility and its median TPS for the day hovered around 1500. The chain with the second highest median TPS for the day, BNB, registered around 180 TPS.

EXHIBIT 3⁸

That said, performance remains essential for certain RWA use cases — particularly public equities, which could drive significantly higher transaction volumes as major chains are actively developing onchain settlement capabilities. For context, major platforms like DTCC, NASDAQ, and Intercontinental Exchange (ICE) process tens to hundreds of thousands of transactions per second through existing infrastructure today.

As traditional market participants in public equities continue to build blockchain capabilities, multiple higher-performance chains will be a requirement. For example, LayerZero’s new Zero chain (announced February 2026) is projected to have a theoretical limit of 2M TPS per zone. While Zero has not yet publicly demonstrated performance, its heterogeneous, multi-zone architecture looks to exceed performance standards set by other chains. As such, many interviewees noted expectations of a multi-chain landscape to handle all the activity that could come onchain — the work to be done is to ensure that liquidity will not be fragmented across chains, but be able to be seamless and interoperable.

Cost

Notably for RWA use cases, not many viewed cost as a decisive issue. As one executive put it: “If it’s secure and defensible, we’ll pay for it.”

While the cost of operating on a blockchain had been a historical concern for some — often stemming from the inflated gas prices due to congestion — today, institutions are less concerned about the overall magnitude of costs but rather their complexity. Cost can feel negligible when, for a $100 million bond issuance, transaction fees are <$0.20 on Ethereum and <$0.01 on Solana⁹, much smaller on a relative basis than for a P2P payment. Rather, institutions would prefer costs denominated in USD (or Euro, local currency), rather than in native network tokens, with (i) simple USD per transaction fee or (ii) bundled pricing for a maximum number, or unlimited, transactions.

Despite being a lower-priority consideration, some RWAs have particular sensitivity to gas fees — for example, TMMFs have regular and frequent corporate actions such as dividends. Lower fees can allow fund managers to distribute payments daily rather than monthly. Exchanges such as NYSE and NASDAQ facilitate billions of daily transactions¹⁰ and even a fraction of these transactions moving onchain can lead to millions of dollars in costs.

Fees matter not just in cost, but in consistency, as they can spike during network congestion. As shown in EXHIBIT 4, during the January 31 crypto market liquidations, median per-transaction fees on Ethereum L-1 spiked nearly 500x from $0.018 to $8.67 in less than 24 hours (due to gas fees being a reflection of demand). These types of differences can affect the underlying economics for a given use case in times of heavy network traffic.

EXHIBIT 4

III. Ecosystem traction

Economic activity

The most important factor outside of security and reliability is the commercial potential that a blockchain’s ecosystem can support — this can be signaled through a combination of quantitative factors (like TVL, though now carefully scrutinized to filter out volumes that can “overstate” volumes, such as looping) and qualitative dimensions, such as the sophistication of its DeFi applications. As one leader put it:

“My commercial goal is to figure out how to distribute my tokenized fund onchain, who the investors are, what the use cases are, and what is required to drive organic growth.

While we love blockchain and the technology, in reality, adoption will be driven by commercial outcomes, not by technology or operational efficiency.”

Institutions interpret economic activity through a few lenses:

• Deep liquidity: volume of stablecoins, active users, and other signs of secondary liquidity can ensure an asset can have proper mobility
• Organic activity: DEX trading (e.g., Uniswap, Jupiter), lending and borrowing protocols and vaults (e.g., Morpho, Aave) can be positive signals for institutional issuers that RWAs can generate utility for actors onchain
• Institutional and enterprise customers: serving populations that are already operating on a given chain — end client chain preferences can inform where institutions build or launch onchain assets (e.g., if the bank is offering a tokenization platform)

Apollo’s tokenized diversified credit fund, ACRED, illustrates this approach. ACRED currently distributes its tokens on 7 different public chains¹¹, with the leading chain holding ~30% of the total ~$130M fund asset value (at time of writing). In addition, through a collaboration with Morpho (a leading lending protocol on Ethereum with ~$5.75B TVL), ACRED can be used as collateral to borrow stablecoins and generate yields through leveraged looping — creating demand effects where borrowing activity generates yield, attracts depositor liquidity, and scales distribution in ways a standalone tokenized fund cannot. Institutions will continue to push into creative ways to drive secondary liquidity for regulated RWAs on public chains.

Institutional adoption

Institutional adoption can be a powerful signal for other institutions determining their digital assets strategy and approach — this dynamic can favor incumbents that have demonstrated performance. Additionally, the presence of regulated custodians, transfer agents, reliable onchain data providers, and audit firms create a network of trust and assurance. For an institution, the decision to use a specific blockchain is significantly de-risked if “traditional” partners, such as State Street or BNY Mellon, are offering services onchain.

The growth of institutional adoption is not only reserved for tokenized products, but also applies to traditional products like ETFs and ETPs. As shown in EXHIBIT 5, major asset managers are enabling crypto price and staking exposure to traditional investors seeking alternative investments but wanting to manage via typical accounts. Most notably, BlackRock’s Bitcoin Trust ETF (IBIT), has been one of the firm’s most successful ETF products to date, currently with ~$51B¹² in net assets and nearly $100B in net assets at its peak in October 2025.

EXHIBIT 5

Developer engagement

The footprint of developer activity plays a practical role as a direct indicator of talent onchain. Ethereum’s first-mover advantage created a generation of developers who grew up on EVM programming and the Solidity language. This established greater familiarity with EVM-based tooling and a deeper talent pool that institutions can more readily engage, creating a sizeable mental and technical moat.

Despite EVM’s early dominance, some public chains utilize non-Solidity-based languages. A few examples include Rust (Solana), Move (Aptos), and DAML (Canton), which can introduce a learning curve for in-house developers who began their journey on Solidity. Chains may opt for non-Solidity-based languages for varied reasons — for Canton, DAML is a language designed specifically for modeling financial agreements, appealing directly to a financial institution’s needs for clarity and verifiability. However, the fluency barrier can require Digital Assets teams to make tradeoffs, including whether to upskill for a potential new environment — this “multilingual” demand on developers can impose a tax on internal resources, require engaging third parties, or depend on the Foundations that build low-code tools.

As institutions scale more formal development teams, blockchains have responded through the professionalization of their ecosystem support (e.g., developer onboarding trainings, low-code/API-driven solutions). In addition, new functions such as Ecosystem Development and Developer Relations have emerged, with the focus on driving a deeper and more engaged development community for various onchain apps and protocols, which can ultimately attract institutions seeking distribution.

IV. Soft factors and influence

Brand, reputation & relationships

Brand & reputation
A final point of differentiation is the various “soft factors” that lend themselves to the credibility of a chain. For some institutions, this can begin with an investigation into sentiment and perceptions. Chains with a sustained history of illicit activities, scams, or “rug pulls” are screened out immediately. While the nature of public chains will never prevent all undesirable activity — just like the Internet as a decentralized protocol cannot regulate all activity — it is up to each institution to determine how their brand profile aligns with a given chain, how it has developed over time, and where it markets itself moving forward.

In addition, institutions are looking at the culture of a given chain and the reputation of their foundation — reinforced by demonstrated thought leadership, activity on X, in-person conferences, and strategic talent hires can all serve as positive signals.

Foundation relationships
The level of engagement with institutions from a chain’s foundation can also play a role in their perception. While some chains take a more neutral, “self-serve model” to institutional engagement, others take an active approach to institutional activity growth and business development. Some foundations have business development teams that engage and support institutions through their onchain journey.

Some cited this active approach as an attractive selling point to operate on a given chain. The sense of partnership can be a meaningful differentiator for TradFi as they explore a new, innovative technology offering. Others noted that the professionalism of a chain’s approach mattered, and those able to approach institutions with a strong knowledge of traditional finance were able to quickly build trust.

Ultimately, for institutions, while technological distinguishers are important, choosing a blockchain ultimately has strong parallels to selecting a long-term business partner.

The emerging reality: Multi-chain and use case-specific

If there is one consistent insight from respondents, it is that institutions do not agree on how the future will play out — many anticipate a multi-chain world. Others expect a robust coexistence of private and public infrastructures as interoperability becomes more seamless and reduces liquidity fragmentation, with some chains dominating specific use cases (e.g., for payments, for high-frequency trading).

While the distribution of value across chains remains uncertain, cross-chain interoperability is becoming critical infrastructure. As use cases specialize by chain, the ability to move assets and data securely between ecosystems will shift from convenience to necessity. For tokenized equities alone — if even a fraction of the multi-quadrillion in annual trading volumes that is managed through clearinghouses such as DTCC comes onchain¹³ — there is no clear path currently to supporting those volume levels on a single, global-state blockchain given current technical capabilities. However, interoperability can introduce new operational and security risks (e.g., bridge design, message validation, and incident response), which institutions increasingly treat as part of core chain diligence.

Conclusion: The infrastructure era is underway

Tokenization has moved beyond experimentation. Major global institutions are deploying real assets on public chains, and the decisions they are making today about blockchain infrastructure will shape the architecture of financial markets for years to come.

Our conversations with 20+ Digital Asset leaders reveal that these decisions are pragmatic, risk-first, and increasingly commercial. But they also surface a set of tensions that the industry has not yet resolved:

• Security vs. ecosystem breadth: the chains with the deepest liquidity and developer ecosystems do not always score highest on decentralization or uptime — forcing institutions to weigh network resilience against commercial reach
• Privacy vs. composability: permissioned environments offer the transaction confidentiality institutions require for sensitive use cases, but at the cost of the open composability that drives DeFi integration and distribution
• Standardization vs. specialization: a multi-chain future appears increasingly likely, with different chains optimized for different use cases — but this raises challenging questions about liquidity fragmentation, cross-chain risk, and operational complexity

Each institution has a different view on the tradeoffs it is willing to make, often largely dependent on their use cases and customer bases.

For institutions, the imperative is to build internal evaluation capabilities now — not to pick a winner, but to develop rigorous frameworks and processes for blockchain diligence that account for the unique needs of each use case they are exploring. This means investing in Digital Assets talent, establishing clear governance for chain selection, and stress-testing assumptions about security, privacy, and ecosystem durability before assets go live onchain. We are still in the early days of blockchain — the industry will continue to evolve, and institutions must be prepared to respond to technological and market shifts.

For Layer-1 and Layer-2 chain foundations, the signal from institutions is clear: technical performance is necessary but not sufficient. The chains that are increasingly winning institutional adoption understand the professional, consultative engagement model to guide institutions from pilot to production, offer credible solutions and clarity on privacy and compliance, and demonstrate that their ecosystems can support not just experimentation, but production-grade financial products at scale.

The infrastructure era of blockchains is rapidly approaching — and the choices made today will determine which rails institutional finance run on tomorrow.

Note: All figures (e.g., AUM, market cap, distributed asset values) as of February 2026
¹ BCG x Dfns: “The Future is Onchain” (January 2026)
² RWA.xyz
³ Layer-1 chains refer to the base network protocols of a blockchain, such as Ethereum mainnet or Solana, which process and finalize transactions independently
⁴ RWA.xyz
⁵ The Nakamoto Coefficient measures how decentralized a blockchain is by identifying the minimum number of independent entities (validators, miners, or staking/mining pools) that collectively control enough power to disrupt consensus
⁶ Chainspect “Most Decentralized Blockchains by Nakamoto Coefficient” (February 2026)
⁷ Chainspect “Fastest Blockchains by Transactions Per Second (TPS)” (February 2026)
⁸ Blockworks “Is the bottom in?” (February 2026)
⁹ Chainspect “Blockchain Financials” (February 2026)
¹⁰ BestBrokers “Stock Trading Market Statistics in 2026” (January 2026)
¹¹ Including Ethereum, Solana, Aptos, SEI, Ink, Avalanche, and Polygon (RWA.xyz, February 2026)
¹² As of February 2026 (BlackRock)
¹³ DTCC “DTCC Processes Record Volumes Across Services Amid Market Volatility” (April 2025)

How Global Institutions are Evaluating Blockchain Networks for RWAs was originally published in BCGonTech on Medium, where people are continuing the conversation by highlighting and responding to this story.

By: Stephen Edison, Helen Kondos, Jared Feiger, Glenn Fiedler

For decades, large law firms grew through steady rate increases, leverage models optimized around the billable hour, and headcount expansion. That model now faces structural pressure as AI drives efficiency for both law firms and their clients.

According to BCG’s 2026 Legal AI Survey, 81% of legal professionals expect AI to materially change the law firm business model within three to five years. So far, most law firm respondents are still in the early stages of adoption (Exhibit 1), and only 20% report significant value at scale. The gap between expectation and scaled execution is where competitive advantage will be won or lost.

The Economic Pressure Is Real and Uneven

AI does not impact all legal work equally. The key variables are addressability (how much work is “on the table” for automation) and feasibility (how readily it can be automated given current technology, data, and regulatory constraints). Tasks that are screen based, formulaic, and output focused score high on both dimensions and include work common in document review, due diligence, and standard contract drafting. Work that is in person, judgment heavy, and relationship driven, such as courtroom advocacy, high-stakes negotiation, and strategic counsel, remains more resistant.

As routine work is automated, per-matter revenue on those engagements will compress, forcing firms to handle higher volume to maintain revenue absent other changes. At the same time, firms will shift their work mix toward matters that are harder to automate, increasing competition in these areas. AI also opens new service categories, from structured data products to AI-powered advisory, that were not commercially viable before.

Tools Alone Are Not a Strategy

Most law firms have begun adopting AI tools, with 73% of respondents either holding licenses and doing limited experimentation, or piloting AI for specific tasks. But adoption without workflow and operating model redesign leaves significant value on the table. Among firms with limited AI experimentation or pilots, only 18% report significant value at scale; among those actively redesigning workflows, that figure rises to 33%. The pattern is consistent when broadening the lens: 92% of those redesigning workflows report at least some organizational value from AI, compared with 68% of those in earlier stages.

Sustainable value comes from changing how work gets done, not merely accelerating existing processes. That means rethinking staffing models, incentives, pricing constructs, and performance metrics alongside the technology itself.

AI Creates a Pricing Paradox

Firms that layer AI onto current workflows risk accelerating margin compression instead of preventing it. Law firm pricing has always reflected knowledge, expertise, and brand, but fees are typically billed by the hour. When AI reduces time spent on a matter, the economics of the billable hour come under direct pressure even as the underlying knowledge and expertise remain unchanged.

Clients continue to value judgment, expertise, and the relationship with their outside counsel, but also expect to share in the efficiency gains that AI creates. The pricing pressure concentrates in larger firms with more sophisticated clients. Among law firms with more than 200 attorneys, 57% report clients asking about AI’s impact on fees, compared with 20% of firms with 200 or fewer attorneys. For large firms, these conversations signal the need to get ahead of pricing pressure before it becomes table stakes.

In response, leading firms are experimenting to find the right mix of billable hour, fixed fee, outcome-based, and other pricing arrangements, and rethinking how those models flow through to internal incentives and structures. Some are restructuring incentive models to reward efficiency and margin rather than revenue alone. Others are exploring retainer arrangements that share upside from productivity gains, or building more rigorous cost accounting to understand true matter economics.

AI Adoption Hinges on Security and Trust

Data security and confidentiality (59%) and trust in AI outputs (56%) are the barriers to AI adoption law firm respondents cite most frequently (Exhibit 2).

These findings point to clear priorities. Security and trust are foundational; without confidence in both, adoption stalls regardless of how capable the tools are. Addressing these barriers requires clear governance and investment in the foundations that make AI reliable in daily work. Security concerns require rigorous vendor diligence, access controls, and clear communication on how data is protected. On trust, firms should pilot tools with structured programs that measure and improve accuracy, building confidence incrementally through demonstrated results.

Firms Face a Strategic Choice on Talent Model

AI forces a fundamental question about the talent pyramid. Firms can maintain the current structure and use productivity gains to handle greater volume with the same headcount, or compress the pyramid and deliver existing work with leaner teams. Most will pursue some combination, and the right answer will differ by practice area and client base. But in either scenario, the composition of teams will change.

Associates will spend less time on tasks AI can handle and more on judgment, and client interaction. Partners will need to manage hybrid teams that include AI agents alongside people. New roles will emerge: professionals who understand both legal practice and the technology that is reshaping it. Firms must invest in upskilling across all levels, while building the in-house technical talent necessary to support AI-native delivery at scale.

Data and Institutional Knowledge Become the Durable Moat

As AI tools become widely available, data and institutional knowledge emerge as the primary sources of sustainable advantage. A firm that systematically structures its work product, precedents, deal data, operational information, and repeatable workflows creates an AI system that learns from the firm’s own experience, unlocking insights competitors cannot replicate.

Much of what makes a firm distinctive, however, lives not in documents but in the tacit and procedural knowledge of its professionals: how experienced lawyers sequence a deal, which judgment calls matter at each stage, and how teams navigate recurring complexities. Firms that find ways to capture and codify that knowledge into structured, machine-readable form will have an advantage that off-the-shelf tools cannot replicate. Knowledge management evolves from overhead into a core enabler of AI-first workflows.

The Competitive Landscape Is Shifting

The competitive environment is shifting on two fronts.

On the demand side, corporate legal departments are building internal capabilities that reduce reliance on outside counsel for standardized work. As the same AI tools become available to both firms and their clients, more routine matters will be handled in-house. This is not a new trend, but AI accelerates it significantly.

On the supply side, new competitors are emerging with aggressive approaches to both technology and talent. Alternative legal service providers and AI-native law firms are deploying AI-first delivery from the ground up, increasingly using agentic systems that execute multi-step workflows autonomously. These entrants are not constrained by legacy leverage structures or billable hour economics, and some are already gaining traction with the world’s largest corporations.

What Leaders Should Do Now

Winning firms are pursuing five priorities in parallel:

• Build an enterprise AI program. Establish executive sponsorship, practice area champions, and structured change management to move AI from experimentation to embedded capability. Redesign workflows end to end, and track metrics that measure whether adoption translates into productivity and quality gains.
• Invest in talent. Upskill associates and partners to manage hybrid teams that include AI agents, and build the in-house technical capabilities required to scale AI-native delivery.
• Reshape the operating model. Align leverage, staffing, incentives, and pricing with AI-enabled ways of working.
• Optimize operations. Apply AI-driven efficiency to the firm’s internal functions to unlock cost savings that fund broader transformation and build the operational infrastructure that AI-native delivery requires.
• Build the data advantage. Turn institutional knowledge and work product into proprietary assets that deliver differentiated client outcomes. This is the moat that will matter most as off-the-shelf tools proliferate.

AI is reshaping how law firms create value, price services, develop talent, and compete. The firms that move decisively will take share, attract talent, and establish positions that compound over time. The next era of legal leadership will be defined not by who adopts AI first, but by who reshapes their business model fastest to capture its value.

AI Is Forcing a Business Model Reset in Law Firms was originally published in BCGonTech on Medium, where people are continuing the conversation by highlighting and responding to this story.

In 2024 and 2025, M&A in the Human Capital Management (HCM) sector rebounded from the 2023 slowdown. The number of M&A transactions in HCM is now nearing historic highs seen in 2021 (see Exhibit 1). This growth is largely driven by five observable archetypes: AI/automation, demand for E2E HCM platforms, industry consolidation, a continued push for global expansion, and portfolio diversification from top HCM platforms. As smaller, standalone HCM firms try to recover from the challenging funding environment of the past few years, they may benefit from a market ripe for further consolidation.

In this post, we will explain how HCM players are using M&A strategies to bolster and broaden their capabilities and retain customers. And, above all, how deals can drive global organizational growth in a rapidly evolving HCM landscape. As noted in Exhibit 1, a lot of 2024 and 2025 M&A activity focused on talent acquisition and compensation & payroll, with additional momentum in AI-enabled solutions outside the core HCM stack. This underscores how vital attracting and retaining top talent is to key growth imperatives while automating core HR operations.

Based on our experience, leading HCM players are leveraging M&A with some combination of these key objectives (see Exhibit 2):

1. Adding AI / automation capabilities

• Growth Impact and Rationale: HCM vendors are currently navigating a critical strategic decision, driven by the urgency to integrate transformative AI to redefine employee experience. Major players may choose to buy best-in-class specialized AI companies to accelerate time-to-market and secure talent and IP, while others may opt to build in-house to ensure architectural fit and long-term control, recognizing the longer timelines involved.
• Example: Workday’s acquisitions of AI-first companies, Sana and Paradox, in the second half of 2025 are intended to give the company a significant competitive stride across their knowledge, learning, and talent acquisition offerings. These moves aim to solidify Workday’s position as a market leader by rapidly integrating specialized, proven, and high-impact AI capabilities that would have taken considerably longer to build in-house.
• Implication: To continue growing amidst shifting customer budgets and priorities, HCM players are moving quickly to add AI / Agentic capabilities, which can often be done faster via acquisition vs. in-house builds

2. Building E2E platforms across the full HCM stack of offerings

• Growth Impact and Rationale: As leading HCM players look to maintain their growth rates, they’re adding functionality across the HCM stack to unlock more cross-sell opportunities and drive retention with a mix of home-grown offerings (e.g., SAP Time Management) and external acquisitions
• Example: In October 2024, payroll and HR giant ADP acquired Workforce Software, enhancing ADP’s HCM offerings by integrating advanced workforce management capabilities for larger enterprises — see Exhibit 2 below for more examples from other HCM leaders like UKG, Workday, and Dayforce
• Implication: HCM players are quickly bolstering their offerings across the full HCM stack and making acquisitions that augment their work. The growth in M&A activity will place more pressure on standalone players and create competition among leading HCM platforms as M&A deals put them all into more intense competition with one another

3. Horizontal consolidation

• Growth Impact and Rationale: Acquiring direct competitors builds competitive advantage through scale, scope and capabilities and/or protection from disruption or threats
• Example: In April 2025, Paychex acquired Paycor, uniting two HCM and payroll platforms serving clients across the U.S. Paychex’s acquisition aims to strengthen their position in upmarket, where Paycor had built significant traction
• Implication: Consolidation immediately increases market share and pricing power and enables the combined entity to create a more profitable and defensible market position

4. Expanding geographically and localizing

• Growth Impact and Rationale: M&A unlocks quick entry into new markets by delivering localized solutions that comply with regional labor laws and standards as well as fully prepared go-to-market teams in the region
• Example: HR Path’s acquisition of GroupeX Solutions in April 2024 significantly accelerated the French company’s geographic expansion of its consultancy and HRIS solutions by strengthening its presence in the North American market
• Implication: Regulatory complexities are one of the largest hurdles for firms expanding globally — M&A helps firms directly scale with products that are already localized & trusted by key actors in the target markets

5. Strategic diversification to HCM adjacencies

• Growth Impact and Rationale: Acquiring HCM-adjacent companies (e.g., that benefit from close integration with HCM stack) can expand into higher margin or growth industries where HCM players have some right-to-win
• Example: Deel’s 2024 acquisition of Hofy, a leading device lifecycle management company, vertically integrated the complex physical and digital IT onboarding experience into Deel’s environment.
• Implication: Creating a new pillar of business allows core HCM players to significantly enlarge their total addressable market (TAM) and additionally enables a stickier primary platform, one more indispensable to the client’s broader enterprise operations.

Outlook and Implications

In this fast-evolving space, there are different sets of imperatives for potential acquirers and acquisition targets. For those looking to make acquisitions:

1. Be clear eyed about your goals for M&A — the competing priorities of (a) adding AI capabilities, (b) fleshing out E2E HCM offering, and (c) expanding into new geos can lead to very different shortlists of potential targets. Focus is vital
2. Be open to opportunistic plays, especially with cash-strapped standalone providers that are feeling increased competition from HCM platforms expanding into their core markets, especially through growing E2E offerings
3. Think ahead to execution — you team’s ability to realize revenue synergies (e.g., cross-selling on Day 1) and successfully integrate products and culture will make or break many deals in HCM

For those looking to be acquired, there’s a parallel set of considerations for maximizing the exit. Given the above, acquirers are often looking for profitable growth in specific geographies and proven ability to sell solutions to clients as both standalone and as part of a broader HCM offering. Leaders of potential acquisitions can help demonstrate their attractiveness by proactively partnering with would-be acquirers and successfully co-selling with them.

With HCM M&A activity continuing to accelerate, these considerations can help both HCM buyers and sellers get the most out of the continuing consolidation.

Authors: Carl Lonnberg, Andrew Foster, Anmol Malhotra, Mohak Mehta, Vishwas Sugla, Liz Klein

M&A as Growth Strategy for HCM was originally published in BCGonTech on Medium, where people are continuing the conversation by highlighting and responding to this story.

By Ben Feldman, Vikram Sivakumar, Clark O’Niell, and Juli Cho

Insights from BCG’s 2025 GenAI in SDLC survey of 500 C-suite technology leaders

GenAI has moved past the experimentation phase in the software development lifecycle (SDLC) and is materially shaping how software gets built. Organizations are now realizing tangible impact — top decile performers report over 30% in productivity gains. Leaders at the forefront with bold ambitions are going beyond to achieve up to 2x performance improvements — indicating an accelerating trajectory that signals far greater gains ahead.

BCG’s latest GenAI in SDLC survey sheds light on what top performers are doing differently to capture value — and where others are struggling. The data points to not tech stacks, company size, or industry as barriers, but to a clear leadership mandate — embed SDLC-specific agents/copilots (not just chatbots and assistants), broaden GenAI usage across SDLC stages (not just coding and development), and invest in fundamental behavior change by moving past launch events and townhalls to redesigning SDLC processes, resetting expectations and accountabilities, providing continuous coaching and enablement, and dedicating time to intensive learning events.

Survey methodology

In November 2025, we surveyed approximately 500 senior technology leaders (e.g., CTO, CPO, CPTO, CIO) who oversee scaled software development organizations (i.e., teams of 50+ product and engineering personnel focused on building proprietary software). The survey focused on four areas: how GenAI is being used across SDLC stages, buying patterns, performance improvements, and the organizational enablers behind those results. The survey was conducted by BCG in collaboration with GLG, an insight network that provides access to expert perspectives.

Survey respondents were located across North America (50%), Europe (30%), and APAC (20%) and distributed across technology, media and telecommunications (~25%), financial institutions (~25%), consumer (~10%), industrial goods (~10%), healthcare (~10%), energy (~5%), and insurance (~5%). ~35% of respondents were large (over $1 billion in annual revenue) and ~45% were midsized (over $100 million but less than $1 billion in annual revenue).

To track trends over time, we compared results to a comparable survey conducted in January 2025 with 100 CIOs/CTOs.

Explore the full study.

Leaders are seeing step-change improvements — proving GenAI is materially shaping how software gets built

GenAI is reshaping software development in ways that are delivering real, measurable productivity gains. Most organizations (~70%) see at least 10% productivity gains, but what’s more striking is the level of impact achieved by the top performers. The top decile performers report >30% productivity gains — and leaders at the forefront with bold ambitions are on track to achieve 2x improvements, indicating an accelerating trajectory with greater gains ahead. This tells us that the goalposts for impact are moving, and teams that declared “mission accomplished” 6–12 months ago with 10–20% gains need to reconsider their approach.

Most organizations are leaving value on the table

Most organizations continue to use general-purpose AI assistants that are not specific to SDLC (e.g., ChatGPT, Gemini, Claude). For example, even in development and coding, where tooling investments are concentrated (with average 2.8 GenAI coding tools deployed), close to 50% of tools deployed are general-purpose AI assistants.

Beyond coding and development, penetration remains low — under 10% of respondents deployed GenAI tools for operations and support, legacy code modernization, testing & QA, and monitoring & observability — indicating that there is more value to capture by embedding GenAI across the SDLC.

What top performers do differently

Top performers don’t rely on modern tech stacks or elite engineering workforces; nor are they all tech companies or small, nimble start-ups. We see leaders emerge across all industries, company size bands, and types of tech stack (see “What we will be watching in 2026” section).

What sets top performers apart are deliberate choices made by executive leadership:

1. Embed SDLC-specific agents/copilots — shifting away from general-purpose AI assistants

2. Scale GenAI across SDLC stages — not just in coding and development

3. Fundamentally change how teams work — moving beyond just launch events and townhalls to redesigning SDLC processes, resetting expectations and accountabilities, providing continuous coaching and enablement, and dedicating time to intensive learning events

Companies don’t need to pay down their tech debt or have structural advantages to reap the benefits from GenAI. Leadership matters more.

1. They embed SDLC-specific agents/copilots — shifting away from general-purpose AI assistants

Top performers embed SDLC-specific agents/copilots, specifically from AI-native startups or frontier labs — going beyond chat or code completion. This contrasts with companies that rely on general-purpose AI assistants that are not specific to SDLC.

For example, in coding and development, almost half of the users of SDLC-specific agents/copilots from AI-natives / frontier labs reported more than 20% productivity gains (see Exhibit 3).

2. They scale GenAI across SDLC stages — not just in coding and development

Top performers are embedding GenAI across multiple SDLC stages — nearly half of respondents seeing top decile productivity gains deploy GenAI tools across 4 or more SDLC stages and vast majority (~90%) across 3 or more — suggesting that impact can compound when GenAI is deployed and scaled across the SDLC rather than as isolated point solutions (see Exhibit 4).

3. They fundamentally change how teams work — moving beyond just launch events and townhalls

Top performers fundamentally change how teams work — moving beyond just launch events and town halls. Organizations seeing top decile productivity gains are ~1.2–1.4x more likely to invest in change management activities — indicating that they treat GenAI as a behavioral and operational shift, not just a technology rollout.

We see the following key patterns (see Exhibit 5):

1. Organizations underinvest in change management overall: Only ~40–50% of respondents report substantial or moderate investment in change management activities.

2. Executive visibility, onboarding, & launch events: All respondents invested similarly in these activities, suggesting that they are “table stakes” that can be deployed across various companies.

3. Intensive learning events & ongoing enablement: Top performers are ~1.2x more likely to invest in activities like hackathons, office hours, and ongoing coaching — suggesting that they are going beyond one-time activities such as townhalls and kickoff event and shifting toward continuous learning.

4. Redesigning ways of working & manager activation: Top performers are also ~1.2x more likely to invest in rewiring SDLC rhythms, rituals, and workflows to incorporate GenAI — reinforced with equipping managers to reset accountabilities, goals, and expectations for their teams.

5. Performance management integration & hiring criteria updates: Top performers are also more likely to change performance management and hiring criteria to match their GenAI ambition. However, just as many top performers reported investing not at all in performance management (~26% selected substantial investment vs. ~23% in no investment), reflecting that these are complex levers that may be deployed selectively, based on company context.

The real bottleneck: process and people, not tools

Companies across the board are struggling with processes and people when it comes to realizing impact — not tool performance or maturity.

IT and cyber approvals are cited as the most time-consuming part of the adoption and implementation process (~27%) (though once approved, half deploy tools to end users within three months) — suggesting that slow organizational processes are a critical bottleneck to adoption.

However, the bottlenecks don’t end after approvals and deployment. Organizations report that the biggest challenges to drive adoption and realize impact are change management — specifically lack of upskilling time, change inertia, and lack of training programs — followed by lack of trust and low output quality. These concerns were ranked ahead of four additional challenges provided — cultural barriers, non-compatible technology stacks, unclear ROI, and lack of management support — which were ranked by less than 25% of respondents.

What leaders can do to unlock impact

Technology is moving quickly and a growing number of organizations are seeing substantial impact — but getting to that impact requires leaders to act above and beyond deploying tools.

· Set bold, forward-looking targets: Leaders should set bold, forward-looking targets of >50% productivity and quality gains. Top performers are already delivering 40–50% productivity improvements — leaders should avoid anchoring to past benchmarks.

· Remove friction to accelerate rollout: Process and people remain the #1 bottleneck. Leaders can simplify approval processes, set expectations for speed through clear timelines, and enable teams to test and scale rapidly. The leadership challenge is to remove unnecessary bottlenecks, starting with time-consuming processes.

· Drive usage and impact, beyond tool deployment: Tools don’t deliver outcomes alone. Leaders should empower teams to fundamentally change how they work — moving beyond launch events and townhalls to redesigning SDLC processes, resetting expectations and accountabilities, providing continuous coaching and enablement, and dedicating time to intensive learning events.

What we will be watching in 2026

We’ll be watching three themes closely in 2026, which may shape the trajectory of GenAI in software development:

1. Top performers emerge across all tech stacks, industries, and company sizes
2. The next frontier — measurement, legacy modernization, and testing & QA
3. As experimentation winds down, what will define winning tools?

The below section highlights key observations and additional survey findings.

1. Top performers emerge across all tech stacks, industries, and company sizes

As expected, we do see some correlation between modern architecture, industry, and company size and impact. However, what’s more striking is that leaders emerge across every aspect — indicating that they are not barriers to impact.

• Architecture: Modern architecture can accelerate, but legacy tech is not a blocker. ~15% of organizations working on service-oriented codebases and cloud-native codebases see top decile productivity gains, compared to ~8% of those working on more monolithic architectures — signaling that success depends more on execution and enablement than underlying tech.
• Industry: Organizations achieve top decile productivity gains in every industry. In addition, all industries see ~70%+ respondents with >10% productivity gains. As expected, some industry contrasts are also emerging. Technology, media, and telecommunications companies lead (~17% report top decile productivity gains), followed by consumer (~12% report top decile productivity gains) and financial institutions (~11% report top decile productivity gains).
• Company size: Smaller firms move more quickly — but impact is seen across all size bands. Though smaller firms see more productivity gains (15% of $25–100M firms achieved top decile productivity gains) than large enterprises (8% of $10B+ firms report top decile productivity gains), uplift is seen across all size bands.

2. The next frontier — measurement, legacy modernization, and testing & QA

Despite extensive conversations about GenAI impact across the SDLC, usage and impact are still in the early innings. We will be watching how usage scales across SDLC stages, with a specific focus on three categories.

• The measurement gap continues: Interest in productivity measurement is growing as teams scale their GenAI investments. It is the fourth largest tool category and 76% of orgs are adjusting developer allocation. Yet firms continue to struggle with measuring developer capacity allocation and only <50% can quantify impact from GenAI confidently. Most who report having something in place rely on self-built dashboards (~25%), while <10% use dedicated productivity measurement tools such as DX, (Developer Intelligence Platform), Jellyfish, and Faros. While productivity measurement tools are unlikely to be a barrier to GenAI adoption, organizations seem to be increasing their focus on measurement as they invest in GenAI tooling.
• Legacy modernization lagging expectations: Despite extensive discussion around GenAI’s potential for legacy code modernization, investments remain low, with <10% of orgs leveraging some of the dedicated tooling that is coming to market focused on this specific challenge — a gap we’ll track as AI-native refactoring and migration capabilities mature.
• Testing & QA may be the next breakout: A year ago, CIOs called out testing & debugging as the number one area GenAI could support — but testing & QA remains underpenetrated, with only 5% of orgs deploying a GenAI tool. As the initial deployments and implementations in coding and development mature, organizations may look to testing & QA as the next breakout.

3. As experimentation winds down, what will define winning tools?

Survey findings indicate that many organizations are still in an experimentation phase that is beginning to wind down. As the market matures and usage consolidates, winning tools will emerge. What characteristics will define winning tools?

• From experimentation to consolidation: Today, firms are using an average of 2.8 coding tools — indicating experimentation is still underway and many organizations have not (yet?) decided to make a bet on one or two core tools. Yet firms are beginning to worry about switching costs (70% ranked as a top challenge when selecting a development & coding tool). As switching costs rise, will experimentation slow — and where will usage consolidate?
• Competitive dynamics — Enterprise vendors vs. AI-native challengers: Buyers tend to first deploy existing vendor suites (e.g., GitHub Copilot, Atlassian Rovo, Figma Make), but buyers are also starting to experiment (~60% of GitHub Copilot users add 2+ other GenAI coding tools). Some users of tools from AI-natives or frontier labs report higher impact. Will enterprise incumbents close the impact gap, or will AI-native challengers continue to expand share?
• Rising prominence of DevEx: Survey data showed that buyers prioritize performance when selecting coding tools, followed by security and developer experience. However, in separate head-to-head benchmarking of coding tools, performance varied little when held constant for model and task. Instead, DevEx, usability, and workflow fit drove differences. Given that survey respondents reported higher productivity gains when using some coding tools over others, we’ll be tracking whether DevEx — rather than inherent tool performance — emerges as a key driver of adoption and value.

Conclusion: Leadership mandate in GenAI in SDLC

The impact of GenAI on the SDLC is no longer theoretical — organizations are already achieving real and measurable impact. Technology is ready — but leadership determines who captures the value. The teams winning today are embedding SDLC-specific agents/copilots, scaling AI across the SDLC, and investing deeply in the people and processes to drive adoption beyond tool deployment. The moment for decisive action is now — companies that lead will redefine how software is built; others will risk falling behind as pace of change accelerates.

GenAI in SDLC: What top performers do differently — and what leaders should do next was originally published in BCGonTech on Medium, where people are continuing the conversation by highlighting and responding to this story.

By Ben Feldman, Vikram Sivakumar, Clark O’Niell, and Juli Cho

Insights from BCG’s 2025 GenAI in SDLC survey of 500 C-suite technology leaders

GenAI is already changing how software is built and run — but adoption within software development teams remains uneven. Most organizations have deployed tools in development & coding. However, rollout of tools covering testing, operations & support, and legacy code modernization all lag far behind — representing significant untapped potential.

At the same time, early adopters are realizing real impact — top decile performers see >30% productivity gains and the number of organizations reporting productivity and quality gains has climbed since early 2025. Leaders at the forefront with bold ambitions are going beyond to achieve up to 2x performance improvements — indicating an accelerating trajectory that signals far greater gains ahead.

Our findings from BCG’s latest GenAI in Software Development Lifecycle (SDLC) survey shed light on the state of the market as SDLC-specific agents/copilots from AI-natives and frontier labs gain momentum, buyers look for performance in an evolving vendor landscape, and early adopters accelerate to realize impact.

Survey methodology

In November 2025, we surveyed approximately 500 senior technology leaders (e.g., CTO, CPO, CPTO, CIO) who oversee scaled software development organizations (i.e., teams of 50+ product and engineering personnel focused on building proprietary software). The survey focused on four areas: how GenAI is being used across SDLC stages, buying patterns, performance improvements, and the organizational enablers behind those results. The survey was conducted by BCG in collaboration with GLG, an insight network that provides access to expert perspectives.

Survey respondents were located across North America (50%), Europe (30%), and APAC (20%) and distributed across technology, media and telecommunications (~25%), financial institutions (~25%), consumer (~10%), industrial goods (~10%), healthcare (~10%), energy (~5%), and insurance (~5%). ~35% of respondents were large (over $1 billion in annual revenue) and ~45% were midsized (over $100 million but less than $1 billion in annual revenue).

To track trends over time, we compared results to a comparable survey conducted in January 2025 with 100 CIOs/CTOs.

Explore the full study.

Across the SDLC, GenAI usage remains limited and general-purpose AI assistants dominate, signaling white space

Today, GenAI usage is limited across the SDLC — dominated by general-purpose, chat-based AI assistants and largely concentrated in development & coding — with rest of the market still underpenetrated (see Exhibit 1). We see two important patterns:

1. General-purpose AI assistants dominate the SDLC: Across SDLC, general-purpose AI assistants that are not specific to the SDLC (e.g., ChatGPT, Gemini, Claude) dominate. This applies even in development and coding, where we see heaviest GenAI tool usage (e.g., 44% of tools in development & coding are general-purpose AI assistants) — indicating opportunity to leverage SDLC-specific agents/copilots for higher impact.
2. Market is underpenetrated across the SDLC: Organizations average 2.8 GenAI tools deployed for development & coding, though almost half of those tools are general-purpose AI assistants not specific to the SDLC. Moreover, only <10% of organizations deploy GenAI tools in operations & support, legacy code modernization, knowledge & documentation, testing & QA, and monitoring & observability.

These patterns underscore that GenAI usage is still nascent or largely absent in the majority of SDLC stages. They signal white space, especially in areas like QA, observability, and operations — where pain points are well known but GenAI solutions, outside of general-purpose AI assistants, are only beginning to be utilized.

Companies across industry, size, and tech stacks are seeing meaningful impact — and momentum has accelerated since early 2025

Although GenAI deployment across the SDLC is underpenetrated, that has not stopped best-in-class organizations from driving real impact with GenAI tools (see Exhibit 2). We see three key patterns:

1. More organizations are seeing productivity gains: Approximately 32% of respondents are seeing productivity improvements of >20%, compared to 22% in January 2025.

2. More organizations are seeing quality gains: Approximately 23% of respondents are seeing quality improvements of >20%, compared to 17% in January 2025.

3. Results indicate uplift is possible for companies of different shapes and sizes: Companies achieving top decile productivity gains don’t rely on modern tech stacks or elite engineering workforces; they are neither all tech companies nor small, nimble startups (see “What we will be watching in 2026” section below).

Our findings suggest that a wide variety of organizations are shifting away from early pilots to more mature implementation, leveraging fit-for-purpose tooling — and generating real value.

Buyers prioritize performance — yet struggle to differentiate between tools in an evolving market

Buyers indicate that they care most about performance (e.g., consistency & reliability, core capabilities) when evaluating GenAI tools across SDLC stages, followed by security and usability, ahead of six other selection criteria provided, which include: agentic features, cost, ease of integration, trust & safety controls, customization, and innovation (see Exhibit 3). However, most organizations still default to established vendors with strong market record or current vendors (~60% of respondents).

These findings indicate buyers still struggle to differentiate between tools in a crowded and evolving market — experiencing a strong pull toward current or established vendors. As experimentation slows and usage of GenAI in SDLC becomes more stable, it remains to be seen how buying patterns will emerge (see “What we will be watching in 2026” section below).

General-purpose AI assistants still dominate — but SDLC-specific agents/copilots are growing rapidly with early signs of higher impact

In development & coding, general-purpose AI assistants that are not specific to the SDLC (e.g., ChatGPT, Gemini, Claude) still dominate, followed by SDLC-specific agents/copilots from “traditional” enterprise vendors (e.g., GitHub Copilot). But the market is shifting quickly.

Since January 2025, SDLC-specific agents/copilots from AI-native startups or frontier labs (e.g., Claude Code, OpenAI Codex, Cursor) have gained meaningful traction in a surprisingly short period.

We see three key patterns:

1. General-purpose AI assistants still lead the market: For example, in development & coding, general-purpose tools dominate, followed by SDLC-specific agents/copilots from “traditional” enterprise vendors (e.g., ChatGPT used by ~65%, GitHub Copilot used by ~62%) (see Exhibit 4).

2. SDLC-specific agents/copilots from AI-natives / frontier labs are growing rapidly: Cursor was used by 1% of respondents in early 2025, but used by 22% in November 2025. Claude Code and OpenAI Codex were both released mid-2025, yet reached ~22% and ~12% respectively by November 2025.

3. Early signs suggest that users of SDLC-specific agents/copilots from AI-natives / frontier labs see higher productivity gains: ~50% of users of such tools see >20% productivity gains — suggesting outsized impact compared to general-purpose AI assistants (see Exhibit 5).

If this pattern continues, we may see a steady migration toward SDLC-specific, AI-native tooling that are tightly integrated into development workflows and tuned for specific tasks (see “What we will be watching in 2026” section below).

Success isn’t dependent on tech stack, industry, or company size — top performers go beyond tool selection and invest in fundamental behavioral shift

Driving adoption and realizing impact goes beyond simply picking a tool and providing licenses to employees. Instead, the challenges are organizational — and organizations are increasingly recognizing change management as a critical barrier to adoption (see Exhibit 6).

1. Change inertia & upskilling time are the top 2 challenges: Organizations are increasingly recognizing change management as a barrier to adoption. Specifically, upskilling is a bigger concern compared to early 2025, with ~22% of respondents ranking it as the top concern, compared to ~8% in January 2025. This reflects a growing acknowledgement among organizations that deployment alone is not sufficient.
2. Companies have more confidence in the ROI of GenAI tools compared to early 2025: Though unclear ROI was the top challenge to adoption in early 2025 (~26% ranked as top concern), by end of 2025, it was no longer even in the top five (~5% ranked as top concern) — consistent with more companies seeing performance gains from deploying tools.
3. Cultural barriers are less of a concern: Similarly, cultural barriers are considered less challenging compared to early 2025 — suggesting readiness to adopt AI-native tooling among product / engineering personnel, provided with the right upskilling and change management support.

This is reflected in what sets top performers apart. Top performers acknowledge and address these concerns by investing heavily in change management. They don’t just stop at townhalls and launch events — they fundamentally change how teams work, indicating that they see GenAI as a behavioral and operational shift, not just a technology rollout.

Top performers (organizations reporting top decile productivity gains of >30%) specifically report investing more in the below (see Exhibit 7).

1. Intensive learning events & ongoing enablement: Top performers are ~1.2x more likely to invest in activities like hackathons, office hours, and ongoing coaching — suggesting that they are going beyond one-time activities such as townhalls and kickoff event and shifting toward continuous learning.
2. Redesigning ways of working & manager activation: Top performers are also ~1.2x more likely to invest in rewiring SDLC rhythms, rituals, and workflows to incorporate GenAI — reinforced with equipping managers to reset accountabilities, goals, and expectations for their teams.

These findings suggest that top performers are more likely to recognize change management as a crucial barrier to adoption early and address it proactively, while most organizations are only starting to catch up now.

What we will be watching in 2026

We’ll be watching three themes closely in 2026, which may shape the trajectory of GenAI in software development:

1. Top performers emerge across all tech stacks, industries, and company sizes
2. The next frontier — measurement, legacy modernization, and testing & QA

3. As experimentation winds down, what will define winning tools?

The below section highlights key observations and additional survey findings.

1. Top performers emerge across all tech stacks, industries, and company sizes

As expected, we do see some correlation between modern architecture, industry, and company size and impact. However, what’s more striking is that leaders emerge across every aspect — indicating that they are not barriers to impact.

· Architecture: Modern architecture can accelerate, but legacy tech is not a blocker. ~15% of organizations working on service-oriented codebases and cloud-native codebases see top decile productivity gains, compared to ~8% of those working on more monolithic architectures — signaling that success depends more on execution and enablement than underlying tech.

· Industry: Organizations achieve top decile productivity gains in every industry. In addition, all industries see ~70%+ respondents with >10% productivity gains. As expected, some industry contrasts are also emerging. Technology, media, and telecommunications companies lead (~17% report top decile productivity gains), followed by consumer (~12% report top decile productivity gains) and financial institutions (~11% report top decile productivity gains).

· Company size: Smaller firms move more quickly — but impact is seen across all size bands. Though smaller firms see more productivity gains (15% of $25–100M firms achieved top decile productivity gains) than large enterprises (8% of $10B+ firms report top decile productivity gains), uplift is seen across all size bands.

2. The next frontier — measurement, legacy modernization, and testing & QA

Despite extensive conversations about GenAI impact across the SDLC, usage and impact are still in the early innings. We will be watching how usage scales across SDLC stages, with a specific focus on three categories.

• The measurement gap continues: Interest in productivity measurement is growing as teams scale their GenAI investments. It is the fourth largest tool category and 76% of orgs are adjusting developer allocation. Yet firms continue to struggle with measuring developer capacity allocation and only <50% can quantify impact from GenAI confidently. Most who report having something in place rely on self-built dashboards (~25%), while <10% use dedicated productivity measurement tools such as DX, (Developer Intelligence Platform), Jellyfish, and Faros. While productivity measurement tools are unlikely to be a barrier to GenAI adoption, organizations seem to be increasing their focus on measurement as they invest in GenAI tooling.
• Legacy modernization lagging expectations: Despite extensive discussion around GenAI’s potential for legacy code modernization, investments remain low, with <10% of orgs leveraging some of the dedicated tooling that is coming to market focused on this specific challenge — a gap we’ll track as AI-native refactoring and migration capabilities mature.
• Testing & QA may be the next breakout: A year ago, CIOs called out testing & debugging as the number one area GenAI could support — but testing & QA remains underpenetrated, with only <5% of orgs deploying a GenAI tool. As the initial deployments and implementations in coding and development mature, organizations may look to testing & QA as the next breakout.

3. As experimentation winds down, what will define winning tools?

Survey findings indicate that many organizations are still in an experimentation phase that is beginning to wind down. As the market matures and usage consolidates, winning tools will emerge. What characteristics will define winning tools?

· From experimentation to consolidation: Today, firms are using an average of 2.8 coding tools — indicating experimentation is still underway and many organizations have not (yet?) decided to make a bet on one or two core tools. Yet firms are beginning to worry about switching costs (70% ranked as a top challenge when selecting a development & coding tool). As switching costs rise, will experimentation slow — and where will usage consolidate?

· Competitive dynamics — Enterprise vendors vs. AI-native challengers: Buyers tend to first deploy existing vendor suites (e.g., GitHub Copilot, Atlassian Rovo, Figma Make), but buyers are also starting to experiment (~60% of GitHub Copilot users add 2+ other GenAI coding tools). Some users of tools from AI-natives or frontier labs report higher impact. Will enterprise incumbents close the impact gap, or will AI-native challengers continue to expand share?

· Rising prominence of DevEx: Survey data showed that buyers prioritize performance when selecting coding tools, followed by security and developer experience. However, in separate head-to-head benchmarking of coding tools, performance varied little when held constant for model and task. Instead, DevEx, usability, and workflow fit drove differences. Given that survey respondents reported higher productivity gains when using some coding tools over others, we’ll be tracking whether DevEx — rather than inherent tool performance — emerges as a key driver of adoption and value.

Conclusion: Where market is heading — and what will determine the winners

GenAI usage in SDLC is rapidly moving beyond the experimental phase. Organizations are already capturing meaningful productivity and quality gains, even as large portions of the SDLC remain untouched. Our findings indicate that impact is achievable across IT architectures, industries, and company sizes — supported with the right tools embedded across SDLC and investments in change management.

The foundations to unlock impact are in place: clear productivity gains, growing confidence in ROI, and a rapidly improving tool ecosystem. As 2026 approaches, the question is no longer if GenAI will reshape the SDLC, but which tools will take advantage of the existing white space and capture the market, and which organizations will drive adoption and realize value.

State of GenAI across SDLC: Adoption is still early — but impact is real and accelerating was originally published in BCGonTech on Medium, where people are continuing the conversation by highlighting and responding to this story.