Gavin Baker argues that AI infra is in an extraordinary demand phase, with frontier-model revenue growth, compute scarcity, and pricing power still favoring the infrastructure stack more than application-layer AI. He is bullish on AI, Nvidia, TSMC, and select chip/infrastructure names, but warns that bubbles can still form if wafer supply, capital discipline, or valuation diversity break down.
Watch on YouTube ›Get the market thesis, key claims, assets, contradictions, and follow-up questions from any financial video — then unlock a version personalized to your portfolio, watchlist, and favorite speakers.
Gavin Baker frames the current AI cycle as the most extraordinary moment in the history of capitalism, anchored by the claim that Anthropic added roughly $11B of ARR in a month — more than the combined businesses of Palantir, Snowflake, and Databricks after years of building. He uses that to argue that March/April volatility was not a thesis break for AI investors; it was a chance to add exposure into what he viewed as one of the clearest AI inflections ever, even while the NASDAQ was selling off. A central part of his view is that AI demand is being constrained and shaped by infrastructure bottlenecks, not by lack of product-market fit. He says the market mispriced the impact of the Strait of Hormuz closure on U.S. energy costs relative to Asia/Europe, which improved America’s manufacturing competitiveness and helped keep focus on AI fundamentals. …
Tactically, the setup stays bullish on AI infra while the market remains constrained by power and wafer scarcity; TSMC capacity and frontier revenue prints are the key near-term tells. The main short-term risk is crowding in lower-quality AI names that could unwind violently if sentiment turns.
Over the next few months, the base case is continued outperformance for frontier models, semis, and power-linked infra as usage-based pricing and capex growth reinforce each other. The view weakens if supply expands too quickly, frontier monetization slows, or the market stops paying up for the highest-quality names.
Structurally, Baker sees AI as a durable new industrial regime built around compute, power, wafers, and frontier access. Even if a bubble forms, he thinks the long-run winners will be the companies controlling scarce infrastructure and the models closest to the frontier.
Anthropic’s ARR growth is unprecedented in business history and more extreme than prior software era compounding.
He compares Anthropic’s monthly ARR addition to the lifetime buildouts of major SaaS companies.
March drawdowns in AI were a buying opportunity, not a thesis break, because fundamentals were improving while prices fell.
He distinguishes invalidated positions from temporary underperformance and says March fit the second category.
The Strait of Hormuz closure improved U.S. relative manufacturing competitiveness because U.S. energy inputs fell relative to Asia and Europe.
He argues lower U.S. gas prices versus higher overseas prices helped America.
How did March and April feel from an investing perspective, given the AI and market environment?
He says March felt like a period of underperformance where he disagreed with the selloff and could lean in rather than concede a bad thesis. He describes the AI moment as unprecedented, with Anthropic's growth making the period feel like an extraordinary inflection rather than a normal drawdown.
Why did the closure of the Strait of Hormuz seem relatively good for America?
He argues it improved U.S. relative manufacturing competitiveness because American natural gas prices fell while prices in Europe and Asia rose. He links that to electricity costs, AI inputs, and a broader focus on America's relative position under the Trump administration.
How should investors think about Anthropic and OpenAI valuations relative to past software leaders?
He says the two companies are structurally different, especially because Anthropic appears much more capital efficient and has burned far less money to reach similar revenue. He suggests Anthropic could be worth something like five times unconstrained run-rate revenue if it had all the compute it wanted.
Unlock the full claims, asset map, scores, related transcripts, follow-up questions, and AI chat — shaped around your portfolio, watchlist, favorite speakers, and risks.
