Google is going all in on Anthropic — up to $40B in cash and compute

Google is reportedly preparing to invest up to $40 billion in Anthropic, the AI company behind Claude. That’s not a typo. Forty billion. In cash and compute credits.

Let that sink in for a second. This isn’t just a big check — it’s a strategic land grab. The AI arms race has moved past the phase where throwing a few billion at a startup makes headlines. Now it’s about securing the kind of compute infrastructure that lets you train and run models at a scale most countries can’t match.

And the timing is interesting. This news comes right after Anthropic quietly released Mythos, a model that’s being positioned as a cybersecurity specialist. Unlike the general-purpose Claude, Mythos is narrow and focused — built to detect threats, analyze vulnerabilities, and automate defensive responses. I’ve been playing with it for a few days, and it’s genuinely good at what it does, but it’s not trying to be the next ChatGPT. It’s a scalpel, not a Swiss Army knife.

So why does Google need to pour this much into Anthropic? Two reasons.

First, compute is the new oil. Everyone wants it, and there’s only so much to go around. Google has its own TPUs and data centers, but even they can’t keep up with the demand from internal projects and external partners. By investing in Anthropic, Google locks in priority access to their future models and, more importantly, ensures Anthropic doesn’t get snapped up by a competitor like Microsoft or Amazon.

Second, Google is hedging its bets. They have DeepMind, which is excellent at research and specialized tasks, but Anthropic has a different culture and approach. Safety-first, interpretability-focused, and willing to move slower. That’s valuable in a market where everyone is racing to ship first and fix later. Google wants both horses in the race.

$40 billion is higher than I expected, even for a company with Google’s wallet. For context, that’s more than the entire GDP of some small countries. It tells me that Google sees Anthropic as a long-term bet, not a quick flip. They’re building a moat.

Of course, there are risks. Anthropic has been burning cash fast — training large models isn’t cheap, and Mythos reportedly cost over a billion to develop. The cybersecurity angle is smart, but it’s a crowded space with established players like CrowdStrike and Palo Alto Networks. And let’s not forget that regulatory scrutiny is only getting tighter. A $40B investment in a single AI company will raise eyebrows in Brussels and Washington.

But if you’re Google, you can’t afford to sit this out. The AI landscape is shifting every quarter, and the winners will be those who control the infrastructure, not just the algorithms. This deal is about compute, plain and simple.

Mythos itself is worth a closer look. It’s not going to write poetry or answer trivia questions. It’s designed for one thing: making cybersecurity teams more effective. Early benchmarks show it outperforms GPT-4 on several threat detection tasks, though it struggles with adversarial inputs. That’s a known weakness, and one Anthropic will need to address before enterprises trust it with their crown jewels.

Personally, I think the focus on cybersecurity is smart. It’s a high-value, high-stakes domain where AI can actually make a measurable difference. But I also worry about centralization. If Google and Anthropic control the compute and the models, what happens to smaller players? Open-source alternatives? The AI ecosystem needs diversity, not just scale.

Anyway, that’s my take. $40B is a lot of zeros, but it reflects a simple truth: the AI race is now a resource war, and Google is buying the biggest guns.