Altman Memo Forecasts ‘Rough Vibes’ Due to Resurgent Google
The Takeaway
- OpenAI’s CEO acknowledged Google’s progress in an AI development field that’s been challenging for OpenAI
- Google’s turnaround is a vindication for CEO Sundar Pichai
- OpenAI’s tech lead is narrowing and its cash burn projections have raised questions among investors
OpenAI CEO Sam Altman told colleagues last month that Google’s recent progress in artificial intelligence could “create some temporary economic headwinds for our company,” though he added that OpenAI would emerge ahead.
After OpenAI researchers heard that Google had created a new AI that appears to have leapfrogged OpenAI’s in the way it was developed, Altman said in the memo that “we know we have some work to do but we are catching up fast.” Still, he cautioned employees that “I expect the vibes out there to be rough for a bit.”
The memo foreshadowed Google’s launch this week of Gemini 3, an AI model that software developers say excels in automating tasks related to website and product design as well as in coding—a capability that is one of the most important drivers of revenue at AI firms like OpenAI.
Altman’s comments show that OpenAI’s technological lead over rivals like Google and Anthropic has narrowed. Investors have sunk more than $60 billion into OpenAI, recently valuing it at $500 billion, on the belief it will continue to dominate the market for developing AI that creates content and reasons the way humans do.
An OpenAI spokesperson declined to comment.
That domination is teetering. Anthropic, a four-year-old firm whose founders previously worked at OpenAI, appears poised to generate more revenue than OpenAI this year from selling AI to software developers and businesses through an application programming interface, The Information reported earlier this month. Anthropic’s models specialize in generating computer code based on what customers want to develop, from new apps to updating existing code.
Google, meanwhile, continues to use its search app and other products to promote its Gemini chatbot, which competes with OpenAI’s ChatGPT. To be sure, ChatGPT is significantly ahead of the Gemini chatbot in terms of usage and revenue, but the gap has been shrinking.
“ChatGPT is AI to most people, and I expect that to continue,” Altman said in the memo.
Google’s other advantage is economic. OpenAI is one of the fastest growing businesses in history, going from next to no revenue in 2022 to a projected $13 billion this year. But it also has projected it would burn more than $100 billion in pursuit of human-level AI in the coming years while spending hundreds of billions of dollars to rent servers to do it, meaning it will likely need to raise the same amount in additional capital.
Meanwhile, Google, valued at $3.5 trillion, generated more than $70 billion in free cash flow over the past four quarters alone. While ChatGPT looks poised to take a bite out of Google Search, Google’s financial performance has improved, in part because it also has a booming cloud business that rents out servers to large customers including OpenAI and Anthropic.
The financial disparity between OpenAI and established firms like Google has prompted public market investors to question whether the startup’s unprecedented revenue growth, including projected growth, will be enough to erase concerns about its future cash burn.
Google’s turnaround is a vindication for CEO Sundar Pichai and his decision to merge dueling AI labs within Google as well as to effectively pay $3 billion last year to bring back longtime AI researcher Noam Shazeer, who had left to launch a chatbot startup.
Training Challenges
Altman, in his note, acknowledged that “by all accounts, Google has been doing excellent work recently,” especially on pretraining, the first phase of developing a large language model that can generate text or images. In that phase, AI researchers expose an LLM to data from the web and other sources so it can learn connections between them.
Google’s success with pretraining in particular came as a surprise to many AI researchers, given that OpenAI at times has struggled to eke out gains from pretraining, an issue Google also wrestled with for a while. Those challenges previously prompted OpenAI to focus more on a newer type of AI model, known as reasoning, which uses more processing power to produce better answers.
And before OpenAI launched its GPT-5 model this summer, its employees found the tweaks they made to the model during pretraining worked when the model was smaller in size but stopped working as it grew, The Information previously reported. That suggests OpenAI will have to resolve these pretraining issues to catch up to Google in that field.
Altman last month assured staff that OpenAI would gain ground in the coming months, including with a new LLM, codenamed Shallotpeat. In developing that model, OpenAI aims to fix bugs it has encountered in the pretraining process, according to a person with knowledge of the model.
Altman said he wanted to focus on “very ambitious bets” technologically, even if it meant that OpenAI would get “temporarily behind in the current regime.” They include advancements in using AI to generate data that could train new AI and “posttraining” techniques such as reinforcement learning, which is essentially a way to rate a model’s answers positively or negatively so it can learn to improve them.
He has privately and publicly discussed the company’s bet on automating AI research itself as a way to speed up breakthroughs, including an AI’s ability to surpass humans in everything from energy and biotech research to healthcare.
“We need to stay focused through short-term competitive pressure,” Altman said in the memo. “We have built up enough strength as a company to weather great models shipping elsewhere. But having most of our research team stay focused on really getting to superintelligence is critically important.”
He added: “I don’t want this note to be a downer—we are doing remarkably well as a company…and I expect that to continue.”
He ended by saying that “it sucks that we have to do so many hard things at the same time—the best research lab, the best AI infrastructure company, and the best AI platform/product company—but such is our lot in life. And I wouldn’t trade positions with any other company :)”