The Bear Case Against Large-Language Model Developers

When Axios on Monday reported on Google’s plan to revamp Assistant—its voice assistant that competes with Apple’s Siri—with technology based on large language models, it got us thinking about a trillion-dollar question that has consumed the tech industry for months. Which software companies will benefit most from the LLM boom: upstart LLM providers like OpenAI and the firms that are chasing it; tech giants like Google, Meta Platforms and Microsoft that develop or license the LLMs; or the flurry of startups trying to bring the tech to the everyday lives of consumers and enterprise workers?

At first blush, the biggest incumbents seem well positioned to quickly incorporate LLMs wherever they might be useful. (And we’re hearing great things about Gemini, the suite of LLMs Google is developing for all its products.)

But investors including Spark Capital and Greylock have placed sizable bets on general-purpose LLM developers. All told, more than $12 billion in VC money has gone into about six LLM providers in the past year, according to The Information’s Generative AI Database. To be sure, about $10 billion of that amount is effectively going to OpenAI from Microsoft’s large coffers, but the other investments are nothing to sneeze at.