Four years after GPT-3 made the trajectory obvious, the first wave of AI markets has set. Elad Gil, an early backer of Harvey, Perplexity, and Character.AI, argues that likely winners have now emerged across several categories. In LLMs, the core players are Anthropic, Google, Meta, Microsoft, Mistral, OpenAI, and X.AI. Three or four of those dominate benchmarks, enterprise adoption, and industry spend. Revenue for leading foundation model companies has gone from zero to multiple billions in roughly three years. Cloud AI spend has hit several billion dollars per quarter at major providers.
The structural reason this market locked in fast: scale requires capital at the billions level, and hyperscalers have non-investment motives to fund winners. Amazon backs Anthropic. Google GCP runs Gemini. Microsoft Azure runs OpenAI. These are not purely financial bets. They are cloud consumption plays. That dynamic accelerated consolidation. Newer entrants like SSI and Thinking Machine Labs, built by credible researchers, remain wildcards: either they find novel approaches, keep raising, or become acquisition targets for incumbents buying talent and leverage.
Gil is careful to note that crystallized markets are not closed markets. Stripe launched a decade after PayPal. Facebook launched years after Myspace. The more interesting argument in the full piece is what comes next: a second set of AI markets Gil says look genuinely uncertain today, covering application layers, vertical software, and agent infrastructure. That forward-looking section, combined with his granular breakdown of which application categories have already found dominant players, is why this is worth reading in full rather than taking the summary.
[READ ORIGINAL →]