On April 24, 2026, Alphabet’s Google announced it’s putting up to $40 billion into Anthropic. Here’s how it breaks down: $10 billion in cash right now at a $350 billion valuation, with another $30 billion on the table if Anthropic hits certain performance benchmarks.<\/p>\n
The money is directly tied to a massive buildout of Anthropic’s computing infrastructure. This comes on the heels of their April 6 partnership with Broadcom for multiple gigawatts of next-gen TPU resources set to go live starting in 2027. Word is Google Cloud will be providing around 5 GW of additional capacity over the next five years.<\/p>\n
| Detail<\/th>\n | Information<\/th>\n<\/tr>\n<\/thead>\n |
|---|---|
| Investment commitment<\/td>\n | Up to $40 billion total<\/td>\n<\/tr>\n |
| Immediate cash<\/td>\n | $10 billion<\/td>\n<\/tr>\n |
| Contingent portion<\/td>\n | Up to $30 billion (performance milestones)<\/td>\n<\/tr>\n |
| Valuation<\/td>\n | $350 billion<\/td>\n<\/tr>\n |
| Announced<\/td>\n | April 24, 2026<\/td>\n<\/tr>\n |
| Primary purpose<\/td>\n | Expansion of computing capacity (including ~5 GW Google Cloud allocation over five years)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\nWhat Actually Happened<\/h2>\nAnthropic says Google is putting down the $10 billion now, while the other $30 billion depends on hitting performance targets they haven’t shared publicly. The money will go toward scaling up the infrastructure that powers Claude models. This deepens an existing relationship between the two companies that already included earlier equity stakes and TPU capacity deals.<\/p>\n The timing is interesting. Anthropic’s revenue has been climbing fast \u2014 their annualized run-rate is now above $30 billion, up from roughly $9 billion at the end of 2025. This comes just days after Amazon made a similar (though smaller) strategic bet on Anthropic. And it stacks on top of that April 6, 2026 announcement about additional multi-gigawatt TPU capacity through Google and Broadcom, with hardware expected to come online from 2027 onward.<\/p>\n Why This Matters<\/h2>\nIf you’re running infrastructure or self-hosting, the sheer scale of compute being locked down at the hyperscaler level is what matters here. A deal this big tells you demand for Google Cloud TPUs and the power infrastructure behind them is still exploding, which can squeeze capacity and affect pricing for everyone else trying to run AI workloads on the platform. Best open-weight LLM to self-host<\/a> options start looking even more attractive when frontier labs are securing dedicated gigawatt-scale allocations that regular users can’t touch.<\/p>\n Google’s play here is textbook strategic positioning: they’re supplying the compute Anthropic needs while taking equity in a company that directly competes with their own Gemini models. Most of that cash flows right back to Google Cloud revenue, creating a self-reinforcing loop that locks in workloads and gives Google deeper insight into how TPUs perform at extreme scale. If you’ve been tracking cloud costs, you’ve seen this pattern before. The real question is how fast these effects trickle down to general availability pricing and quota limits.<\/p>\n The Hacker News crowd got it right when they framed this less as rivalry hedging and more as vendor financing with upside: Anthropic gets the capital to grow, Google gets a major committed customer plus equity exposure. For homelab operators and smaller teams, this just reinforces the gap between hyperscaler-backed frontier models and the self-hosted stack you actually control.<\/p>\n Google has been backing Anthropic since at least 2023 and already held an estimated 14% stake before this round. The relationship has always been both collaborative and competitive \u2014 Anthropic runs its biggest workloads on Google Cloud and TPUs while Google keeps developing Gemini through DeepMind. That April 6 TPU partnership with Broadcom already locked in multiple gigawatts of next-gen capacity; this new investment accelerates and funds even more expansion on top of that.<\/p>\n |