AI revenue reached $100 billion in just two years from zero, with Wall Street warning of a potential SaaS market saturation trap as adoption plateaus.

D.A. Davidson analyst Gil Luria captured the AI industry's explosive growth in a single striking observation on The Real Eisman Playbook: "The cumulative run rate of OpenAI and Anthropic right now is clearly above $75 billion." He revised upward in real time: "By the time you include Gemini's revenue and maybe a little bit Meta and xAI, we're above $100 billion of revenue from what was zero a couple of years ago."

Yet buried within this bull case lies a cautionary tale. Luria warned that building a business directly on top of closed foundation models is structurally perilous. "If you build your business on top of a model from either Anthropic or OpenAI and something happens to that model, you're screwed," he cautioned, citing regulatory intervention in Anthropic's Fable model as a case study: companies betting their entire business on a single foundation model face existential risk if that model disappears or becomes unavailable.

The infrastructure layer is capturing the lion's share of this revenue explosion. NVIDIA posted Q1 FY2027 revenue of $81.615 billion, with Data Center alone reaching $75.246 billion and networking up 199% year over year. CEO Jensen Huang framed the moment as "the largest infrastructure expansion in human history," with supply commitments now at $119.0 billion and Q2 guidance pointing to $91.0 billion. Trading around $196.75 with a forward P/E of 23, NVIDIA has become the most reasonably valued mega-cap AI trade, up 13.41% over the past year even as the SaaS layer wobbles.

The same pattern holds across Big Tech. Alphabet posted Q2 revenue of $119.8 billion, with Cloud growth accelerating to 82% and backlog reaching $514 billion. Full-year 2026 CapEx guidance was raised to $195 billion to $205 billion. Microsoft spent $37.5 billion in a single quarter and pushed Microsoft 365 Copilot to 15 million paid seats, up over 160% year over year. Meta Platforms raised 2026 CapEx guidance to $125-145 billion.

But the market is not rewarding scale evenly. Microsoft is down 19.19% year to date despite its AI business revenue crossing a $37 billion annual run rate and crushing expectations. The retail sentiment, captured in a viral Reddit post, cuts to the core concern: "Azure +39%, AI revenue +123%, 4th st. beat, stock down 30%. The market has decided capex is sin."

Luria's warning lands hardest on companies wrapping their applications around closed foundation models. He echoed a structural risk Palantir CEO Alex Karp has raised repeatedly: "If you put your data into Anthropic's model, they have your data. Then they know how your business operates. And if they decide to compete with you, they can compete with you."

Palantir stands as the counterexample—it owns its ontology layer, sits above model providers, and posted Q1 FY2026 U.S. commercial revenue growth of 133%. Yet valuation has delivered its own warning shot. Trading at a trailing P/E of 148 and price-to-sales of 60.36, the stock is down 26% year to date. Polymarket traders assigned a 98% probability of PLTR closing lower that day.

Luria's framework points to a clean bifurcation: infrastructure, led by NVIDIA and sovereign-AI plays deploying models like NVIDIA's Nemotron, is compounding hard cash. Applications built on closed foundation models face platform risk, regulatory risk, and data-leak-to-competitor risk. Prediction markets see the difference: NVDA carries 94.5% to 99.5% probability of holding above $180 through month-end, while PLTR sits in a narrow $108 to $144 band with fading conviction beyond August.