A.I. startups like OpenAI and Anthropic must partner with tech giants like Microsoft and Google. That poses problems

ChatGPT has taken the world by storm. But the artificial-intelligence chatbot has also needed enormous cloud computing resources to do so. The same will go for any rival hoping to overtake it, such as Claude from the startup Anthropic, and other ventures looking to capitalize on “generative A.I.,” which Bill Gates recently called “every bit as important as the PC, as the internet.” While users are enthralled by A.I. programs that can answer questions and generate images in seconds, “the compute costs are eye-watering,” noted Sam Altman, CEO of ChatGPT developer OpenAI, in December. That helps explain why this week OpenAI launched ChatGPT Plus, a $20 monthly subscription that provides faster response times and better access to the A.I. chatbot when it’s otherwise down due to traffic.

Antitrust concerns

But it also points to the need for A.I. startups like OpenAI and Anthropic to partner with the handful of tech giants that can provide them with the enormously expensive cloud computing resources they need to operate. And such arrangements are raising concerns among antitrust regulators. They are “exactly the type of scenario that the Federal Trade Commission has said they’re going to focus on,” William Kovacic, who previously led the FTC, told the Financial Times.

A.I. startups without Big Tech: ‘Impossibly expensive’

Such deals give tech giants insights into the personnel and capabilities at such startups, yet there’s little to stop them from working on similar products themselves. Google subsidiary DeepMind is reportedly preparing to launch Sparrow, another rival to ChatGPT. For startups, jumping from one cloud provider to another is difficult. Many could find themselves essentially locked in to one of them, even when there’s no exclusive agreement. “Some academics that want to move into their own startup, their first conversation is with cloud providers before they even recruit developers because they know it’s impossibly expensive,” Yellow Dog CEO Tom Beese told the FT. “It’s key.” Beese’s firm helps clients switch between cloud services, but he noted he knew of several alliances between A.I. startups and cloud providers formed before products have even been launched. “There is a heightened concern about how the large information services firms are limiting opportunities for new generations of competitors to come forward,” added Kovacic, who now teaches antitrust law at George Washington University. OpenAI is now deeply entwined with Microsoft, which announced last month it will invest billions into the venture. The tech giant also invested $1 billion in 2019, then quietly added another $2 billion in 2021. Anthropic this week announced a $300 million investment from Google that gives the latter a 10% stake in the venture. In late November another generative A.I. startup, Stability AI—users of its tools can generate impressive images from text prompts—announced it had chosen Amazon Web Services as its partner.