Reasons for Diverging AI Infrastructure Valuations

AI multiples are multiplying traditional multiples. AI unicorns command 2.4x higher revenue multiples than their non-AI peers, garnering a median 24x revenue multiple compared to just 10x for traditional unicorns. What's driving sky-high multiples? ↳Foundational capital requirements AI companies require enormous upfront investments for compute infrastructure, model training, and top-tier talent. These capital requirements create natural barriers to entry – companies that can attract the funding, already demonstrate a sustainable competitive advantage. ↳Speed to unicorn status AI unicorns are reaching billion-dollar valuations in roughly half the time of non-AI unicorns – just 3.6 years versus 7 years for traditional unicorns. Youthful AI unicorns are unsurprisingly in earlier revenue stages. ↳Massive tech-quisitions and acqui-hires AI investors are pricing in future monetization opportunities or high-value acquisitions for core tech rather than financial performance, a luxury traditional businesses don't enjoy. ↳Strategic bidding wars Tech giants are paying premium valuations to secure AI capabilities. Microsoft's OpenAI investments, Amazon's Anthropic backing, and Google's various AI partnerships show strategic investors will pay significant premiums for preferential access to AI innovation – driving up valuations across the board. While AI companies are benefiting from a "potential premium", non-AI unicorns are facing heightened scrutiny on metrics that AI companies can defer: clear paths to profitability, sustainable unit economics, efficient capital deployment. They must prove their business models work at scale, and even when they can, they are held to more traditional valuation metrics. Across the broader venture ecosystem, we're witnessing a dual-track funding environment. The AI track is seeing abundant capital, premium valuations, investor competition, and speculation, hope, or FOMO-driven pricing. The non-AI track is seeing capital scarcity, fundamental-based valuations, investor selectivity, and stricter profitability requirements. P.S. We’re building the world’s largest, high-quality private company revenue data set. Check it out at the link in the graphic.

GenAI is the biggest swing factor in SaaS valuations today—doubling multiples for some, leaving others unchanged. Here’s where things stand: 1. SaaS baseline vs. GenAI uplift Most public SaaS names trade around 9× trailing revenue. But companies with a credible GenAI story are seeing multiples in the 17–28× range: • CrowdStrike trades at 28×, with AI powering threat detection and automation. • Snowflake and ServiceNow hover near 17–18×, positioning AI as central to their platform strategy. • Adobe, despite heavy investment in generative tools, has dropped closer to 7× following cautious signals on monetization. The median for AI-forward software companies is around 17×, nearly double the broader SaaS average. 2. Private AI startups are even more aggressively valued Recent deals in the GenAI space are pricing at 23–26× revenue, well above the private SaaS norm of 7–9×. This reflects investor belief in future expansion, even when current usage or monetization is early. 3. Why GenAI adds 8–10 turns The valuation premium isn’t just buzz—it’s grounded in investor conviction around: • Revenue acceleration through new SKUs, pricing power, and AI-led land-and-expand • TAM expansion, transforming point products into full platforms • Scarcity premium, with few scaled GenAI-native players in the market • Margin tailwinds, based on improving inference efficiency and pricing dynamics 4. But the premium is fragile Without clear, monetized AI traction, the multiple deflates quickly. Adobe’s recent dip is a case in point—investors want results, not just vision. In categories like cybersecurity, we’re already seeing a sharp divergence in multiples: those with visible GenAI differentiation are trading 4–5× higher than peers still early in their AI journey. 5. What to watch next • GenAI-specific revenue reporting: More companies will need to show AI’s direct business impact. • Inference cost curves: If infrastructure costs don’t drop fast enough, margin expansion assumptions will need to be revisited. • Platform consolidation: The long-term winners will become the embedded AI layer for enterprise workflows, agents, and copilots—not just feature vendors. Bottom line: GenAI is adding 8–10 full turns to SaaS valuations, but that uplift is fragile. Investors are no longer rewarding potential—they’re rewarding proof.

Help me figure this out... There's a puzzling disconnect in the cloud computing industry today. Major cloud providers consistently claim they're struggling to meet overwhelming demand for AI computing resources, yet their quarterly earnings reports often fall short of Wall Street's expectations. This paradox is becoming increasingly visible. While cloud providers report strong year-over-year growth in AI and cloud segments, these numbers frequently disappoint market analysts anticipating even higher returns. Despite announcing unprecedented capital expenditures for AI infrastructure, often planning 40%+ increases in capital budgets, providers are struggling to demonstrate proportional revenue growth. Investors are questioning these "eye-watering capital expenditures being poured into AI infrastructure." The fundamental concern is that AI remains an expensive research project with significant uncertainty about how the global economy will actually absorb, utilize, and pay for these capabilities at scale. The market seems caught in a contradiction: there's enormous enthusiasm for building AI computing capacity, yet actual implementation and monetization of AI applications remain tepid. Cloud providers may be conflating potential future demand with current market reality, leading to a mismatch between infrastructure investments and immediate revenue generation. While the long-term potential of AI is undeniable, the short-term market dynamics are more complex than providers' public statements indicate. #AI #CloudComputing #EnterpriseTechnology #InvestmentParadox

Everyone's chasing AI models. But smart money is buying something different: Data infrastructure. Here's why Salesforce just dropped $8B on Informatica: 1/ The Hidden AI Crisis: ↳ 80% of AI projects fail ↳ Most never reach production ↳ The culprit? Bad data infrastructure 2/ Enterprise Data Reality: ↳ Customer data locked in Salesforce ↳ Financials trapped in SAP ↳ Marketing data in Adobe ↳ Analytics buried in Mixpanel None of these systems talk to each other. And AI can't function on disconnected data. 3/ Two Competing Visions: The Integrated Stack Players ($6T+ combined market cap): ↳ Microsoft: Azure + Dynamics + OpenAI ↳ Oracle: Database to cloud stack ↳ Google: BigQuery to Vertex AI ↳ Salesforce: MuleSoft + Informatica They want to own your entire data journey. Like private railways of the 1860s. The Open Ecosystem Players ($127B+ combined value): ↳ Databricks: Betting on open standards ↳ Snowflake: Independent data cloud ↳ Elastic: Cross-platform analytics ↳ Confluent: Universal data streaming They're building public infrastructure. More freedom, but more complexity. 4/ Urgent Actions for Companies: First: Audit Your Data ↳ Count your critical data systems ↳ Map existing connections ↳ The mess compounds daily Second: Invest in Data Quality ↳ Clean data makes AI work ↳ Dirty data breaks everything ↳ Choose your stack wisely Remember the cloud wars? AWS started open, then made leaving expensive. Smart enterprises now insist on multi-cloud. The same pattern is emerging in AI infrastructure. Don't get distracted by shiny AI models. Build your data rails first. Your competitive edge depends on it. What do you think about investing money in data? Share below. ♻️ Share this with someone who needs to keep up with tech. ➕ Follow me, Ashley Nicholson, for more tech insights.

Palantir Technologies's 400% surge reveals a strategic shift Wall Street isn't just betting on another AI stock. They're recognizing that application-layer AI companies will capture disproportionate enterprise value over the next 18 months. Our analysis at Global AI Forum shows a pattern emerging. While infrastructure companies build the foundation, application specialists like Palantir solve the enterprise adoption challenge that has stalled $5B+ in AI investments. The strategic disconnect: Most enterprises can access AI capability but cannot implement AI solutions that drive measurable business outcomes. Palantir's AIP platform reducing Citi onboarding from days to seconds isn't just efficiency gain. It's proof that AI applications generate enterprise value faster than AI platforms. Three strategic implications most miss: 1 - Enterprise AI budgets are shifting from infrastructure to implementation. 2 - Government contracts are moving from research to operational deployment. 3 - Retail investor flows into AI applications signal mainstream adoption acceleration. The companies positioning as AI implementation specialists, not just AI developers, capture the next wave of enterprise transformation value. Six months ago, our research predicted this exact market evolution. Application-layer AI companies would outperform platform companies in enterprise adoption cycles. Strategic clarity in AI transformation demands seeing beyond infrastructure to implementation value. Join leaders who recognize the patterns first. Global AI Forum provides the competitive intelligence that shapes strategic positioning decisions.

Everyone's chasing AI. The smartest founders I know are building what AI needs to exist. Last week, a founder pitched me their "AI for sales" startup. They were the 7th AI pitch I'd seen this month so far. I asked one question: "What's your monthly compute bill?" "$287,000. And growing." That's when I showed them where the real opportunity was hiding. The infrastructure paradox: Every AI startup needs: GPUs they can't get Data centers already at capacity Specialized compliance tools Cost optimization they can't build The gold rush is obvious. The shovel shortage? That's where fortunes get made. 1849: Levi Strauss didn't mine gold. He sold jeans to miners. 1990s: Cisco didn't build websites. They sold the routers. 2000s: AWS didn't create apps. They rented the servers. Today's version? Look at what's actually getting funded: GPU scheduling optimization AI model monitoring platforms Specialized cooling systems Compliance and governance tools Not getting funded: "ChatGPT for [insert industry]" What I tell every founder: You don't need to predict which AI company wins. You need to sell to all of them. One founder pivoted from "AI-powered analytics" to "analytics for AI companies." Before: Competing with thousands After: Serving thousands The difference? Every AI company needs infrastructure. Only a few need another competitor. The opportunities hiding in plain sight: Model versioning systems. Compliance frameworks. Data labeling tools. Cost optimization platforms. Inference infrastructure. Edge computing solutions. Boring? Yes. Necessary? Absolutely. Fundable? Ask any VC focused on infrastructure. The smart money isn't just chasing AI. It's building what AI needs to exist. Here's what founders miss: The best businesses in a gold rush aren't the ones finding gold. They're the ones everyone pays on the way to the mountain. Your competition as an AI startup: OpenAI, Anthropic, and thousands of others. Your competition as infrastructure: Usually just spreadsheets and duct tape. The founders getting funded in Q4 won't all be building the future of AI. Many will be building the foundation it runs on. Stop asking "How do I compete with ChatGPT?" Start asking "What does every AI company buy?" That's where the real opportunity is. #VentureCapital #Infrastructure #StartupFunding #AIInvestment #FidelmanCo

AI startups represent about 70% of B2B Series As, up from about 40% in early 2024. On average, AI Series As raise at 40% higher valuation than non-AI companies, a multiple that has been increasing over time. But this is less of a premium than in the public markets. As of January 31, an AI publicly traded software company trades at twice the multiple of a non-AI software company. The major difference between the public markets and the private markets is the relative differential in growth rates between software and AI companies. High-growth early SaaS companies can achieve growth rates similar to those of their AI peers In the public markets, typically, the slower the growth of the company, the less they have been investing in AI. AI companies should trade at a higher multiple for a few reasons : 1. The growth rates of AI companies can be significantly higher. Buyers are more curious about these products. The potential business impact to the buyer is greater as well. 2. The bet is that the AI businesses will tap into labor markets. There’s $1.5t of IT spend. About 40% of it has moved to the cloud. Overall software spend is less than 10% of operating expense for many businesses & labor is multiples greater. 3. AI software reinvents workflows in ways that could enable market share shift from incumbents, in a way that hasn’t been possible since the beginning of the cloud era. There’s truth to the narrative that AI is the new platform & companies that embrace it are rewarded with lower costs of capital.

𝗚𝗼𝗹𝗱𝗺𝗮𝗻 𝗦𝗮𝗰𝗵𝘀 '𝟮𝟯: 𝗚𝗲𝗻𝗔𝗜 𝘁𝗼 𝗮𝗱𝗱 $𝟳𝘁𝗻 𝘁𝗼 𝗴𝗹𝗼𝗯𝗮𝗹 𝗚𝗗𝗣. 𝗚𝗼𝗹𝗱𝗺𝗮𝗻 𝗦𝗮𝗰𝗵𝘀 '𝟮𝟰: 𝗡𝗼𝘁 𝘀𝗼 𝗳𝗮𝘀𝘁. 𝗪𝗲 𝗵𝗮𝘃𝗲 𝗮 $𝟭𝘁𝗻 𝗴𝗮𝗽. (That's at least what the headlines imply. 👇) What's going on beyond the headlines? In their June 2024 research, financial analysts at Goldman Sachs state: "[...] leading tech giants to spend an estimated ~$1tn on capex in coming years. But this spending has little to show for it so far beyond reports of efficiency gains among developers." "[...] to earn an adequate return on the ~$1tn estimated cost of developing and running AI technology, it must be able to solve complex problems, which, it isn’t built to do." That's a pretty stark contrast to April 2023: "[...] they could drive a 7% (or almost $7 trillion) increase in global GDP and lift productivity growth by 1.5 percentage points over a 10-year period." How could they be so far off?! Everyone that I've recently talked to (and who's seen this playbook 5 years ago) agrees: 1) Investment in infrastructure is driving current demand (IaaS/ PaaS) 2) With the infrastructure in place, vendors can build higher-level offerings (SaaS) 3) Large-scale adoption takes time—irrespective of the industry 4) Define your AI strategy today and build your organization's AI muscle If you're looking for a reality check beyond the hype on how to make AI work in your business, let's talk: https://lnkd.in/dAQWzUqQ 𝗪𝗵𝗮𝘁 𝗵𝗲𝗮𝗱𝗹𝗶𝗻𝗲 𝘄𝗶𝗹𝗹 𝘄𝗲 𝘀𝗲𝗲 𝗮 𝘆𝗲𝗮𝗿 𝗳𝗿𝗼𝗺 𝗻𝗼𝘄? (Quotes edited for brevity.) #ArtificialIntelligence #GenerativeAI #Hype #IntelligenceBriefing

Amazon and NVIDIA just redrew the map for AI infrastructure. And it’s happening faster than most people realize... This isn’t just about building more data centers. It’s about energy, sovereignty, and specialization becoming the new gates to scaling AI globally. At the Powering AI Summit, Amazon and Nvidia made it clear: Demand isn’t slowing, it’s shifting to regions and operators that can solve for power, political alignment, and hyperscale-specific designs. Key shifts underway: 1. 50GW of new power needed by 2027 equivalent to 50 nuclear plants (Anthropic estimate). 2. Energy scarcity is now the #1 bottleneck for #AI growth, not land or capital. 3. Sovereign AI stacks are fragmenting the global market, localized compute is becoming mandatory. 4. AI-specific infrastructure (100kW+/rack densities, liquid cooling) is replacing traditional cloud builds. The smartest players, like Amazon and Nvidia, are moving fast into energy-secure, sovereign-aligned regions: #Hyderabad, #HoChiMinhCity, #Querétaro, #AbuDhabi. They’re securing natural gas partnerships, piloting small modular nuclear reactors, and building geopolitical resilience into every new deployment. Meanwhile, investors and operators who keep using old playbooks (cheap land, loose permitting, globalized designs) are already falling behind. The next 5 years will reward those who control energy, sovereignty, and specialization, not just capacity. Are you adapting your AI infrastructure strategy around these realities? Read this week's insights to learn more.

Most calls with co-investors these days include some mention of their interest in AI (and for climatetech / hardtech, specifically on the infrastructure layer). We VCs love a good way to repeat after one another 😊. The Carlyle Group published some good stats: AI-related capex across data centers, GPUs, power, and supporting infrastructure already accounts for more than a third of Q2 U.S. economic growth. Order books for the companies supplying this buildout are still growing +40% YoY. Among Nvidia’s top U.S. customers, capex has grown 1.5x faster than revenue. PP&E now makes up 70% of book value, up from just 20% when cash and securities dominated balance sheets. These are deep shifts in business model and capital structure. The bet is that future AI software and services will drive enough durable revenue to justify this shift. But it’s still unclear whether utilization, depreciation timelines, and revenue per watt or per FLOP will support that thesis. On the software side, the economics are equally unresolved. If inference costs scale with usage and API licensing eats into gross margins, enterprise AI apps may not benefit from the same non-linear operating leverage we’ve come to expect from SaaS. And if value continues to consolidate at the foundational model layer, we may see application-level businesses structurally disadvantaged. I’ve been thinking about what kinds of economic architectures actually support durable returns in AI. The question is whether the business models we’re building around it are as scalable and defensible as the multiples suggest.