The latest sign of the artificial intelligence industry’s relentless pursuit of computing power comes from the expanded agreement between Anthropic and Google, which, like several others struck in recent months, highlights the rise of round-robin arrangements across Silicon Valley. The two companies announced yesterday (October 23) that under the new agreement, Google will provide Anthropic with more than 1 gigawatt of computing power by 2026.
Anthropic noted that the deal is worth “tens of billions of dollars,” but did not provide a specific number. The partnership further deepens the startup’s ties with Google, which has invested about $3 billion in Anthropic and is expected to provide the company with up to 1 million custom artificial intelligence chips called tensor processing units (TPUs).
Such partnerships are increasingly important as leading AI startups expand at breakneck speed. Anthropic currently serves more than 300,000 enterprise customers, and the company says the number of customers with annual revenue of more than $100,000 has grown nearly sevenfold in the past year. “Anthropic and Google have a long-standing partnership, and this latest expansion will help us continue to develop the computing power needed to define the frontier of artificial intelligence,” Anthropic Chief Financial Officer Krishna Rao said in a statement.
Anthropic, founded in 2021 by CEO Dario Amodei and several former OpenAI employees, positions itself as a security-focused alternative for early AI players. The company, best known for its chatbot Claude, was recently valued at $183 billion and is reportedly on track to hit $9 billion in annual revenue.
Despite its close relationship with Google, Anthropic stressed that it remains committed to its “primary training partner” Amazon, which has invested $8 billion in exchange for providing computing through its chips and AI cluster Project Rainier. The company also continues to rely on Nvidia’s GPUs as part of what it calls a “multi-platform approach.” Anthropic said it will continue to invest in additional computing power as demand grows.
Anthropic’s mutually beneficial partnerships with Google and Amazon reflect a broader industry trend toward a growing network of interconnected AI partnerships between model developers and compute providers investing in and buying each other’s technology. OpenAI has been at the forefront of this shift, announcing a series of major deals in recent months, including a deal with AMD for 6 gigawatts of computing power, a deal with Nvidia for 10 gigawatts of computing power, and a five-year partnership with Oracle worth $300 billion.
The growing prevalence of such round-robin arrangements has raised concerns among some in Silicon Valley, recalling the speculative interdependence of the dot-com bubble and its eventual collapse. But unlike that era, today’s AI spending is underpinned by stronger capital and clearer monetization potential, JPMorgan global market strategist Stephanie Arriaga said in a blog post earlier this month.
However, Arriaga warned that such concerns are not unfounded. “The scale of spending is huge and the pace is unprecedented, and some assumptions about return on investment, such as the useful lives of assets, remain in question,” she said. “History reminds us that enthusiasm can trump reality,” she wrote.
