On a Saturday night in mid-September, a senior Google engineer shared some rough news with more than fifty colleagues. Part of the company’s cloud services offering was failing Anthropic, a darling AI startup and key strategic customer, and they’d have to work overtime to fix it.
To repair the faulty part of its service—an underperforming and unstable NVIDIA H100 cluster—Google Cloud leadership initiated a seven-day-per-week sprint for the next month. The downside of not making it work, the senior engineer said, was “too large, for Anthropic (most importantly), for Google Cloud, and for Google,” according to documents reviewed by Big Technology.
One week after Google initiated the sprint, Anthropic announced it landed a new investment of up to $4 billion from Amazon that would make Amazon Web Services its “primary cloud provider for mission-critical workloads.” The Amazon deal was in the works long before Google Cloud’s performance issues and was unrelated, according to a source familiar with the matter, but Google’s sprinting engineers couldn’t have felt good seeing Anthropic Ink make a major deal with the competition. Especially after Google put about $300 million of funding into Anthropic late last year for 10% of the company.
Anthropic is one of the few independent AI research firms that can credibly stand beside OpenAI and the tech giants. A handful of OpenAI expatriates founded the company out of concern that their ex-employer wasn’t as open and safe as it claimed. Earlier this year, Anthropic released Claude, its answer to ChatGPT.
In addition to Claude the bot, Anthropic allows developers to build on top of its AI models, and the training and execution require significant capital and computing. To that end, Anthropic relied on the tech giants to support it, similar to how OpenAI relied on Microsoft. But the agreements are not quite parallel. Even after its Amazon deal, Anthropic will continue to partner with Google, making itself more of a neutral player among AI research houses.
Still, Anthropic’s new funding deal should be a boon to Amazon and leave Google a bit puzzled. For Amazon, it means landing a new strategic partner that can help it advance the research status quo — something it’s been missing despite an eagerness to facilitate building on all models. For Google, the deal leads to a number of questions, especially regarding where it wants to play in AI. As it builds its own AI models with Google DeepMind, will it risk losing AI customers in the cloud? Does it want to prioritize reinventing search with AI, or helping businesses reinvent themselves with it? Can its lower margin divisions like Google Cloud Platform operate with the same urgency as its higher margin search businesses?
Google may successfully be able to do it all at once, but as it looks ahead, it will risk being outflanked by competition with less fraught tradeoffs. Google Cloud’s performance issues with Anthropic, meanwhile, seem to be settling down. But not without forcing engineers into something not often associated with life at Google — weekend work.
