Google Cloud Hits $20B, But the AI Boom Is Straining Its Servers

Google Cloud finally did it. For the first time, the division posted over $20 billion in a single quarter. That’s a big number, and it’s almost entirely thanks to the AI feeding frenzy.

But here’s the part that caught my attention: Google’s own executives said they could have done even more. Capacity constraints. They literally couldn’t spin up enough servers fast enough to satisfy all the demand.

That’s a strange problem to have when you’re one of the three biggest cloud providers on the planet. But it’s also a sign of just how insane the AI infrastructure race has become. Every company with a GPU wants to rent it out, and Google’s TPUs and NVIDIA chips are in high demand.

Revenue hit $20.2 billion for the quarter, up about 35% year-over-year. That’s higher than I expected, honestly. The cloud business has been a slow burn for Google compared to AWS and Azure, but AI is finally giving it a real boost.

Alphabet’s CFO Ruth Porat mentioned during the earnings call that “capacity constraints” limited growth. She didn’t specify exact numbers, but the implication is clear: if Google could have provisioned more compute, that $20B would have been even bigger.

This isn’t a new problem. Google’s been building data centers like crazy, but the supply chain for high-end chips and cooling infrastructure just can’t keep up with demand. Every cloud provider is facing this, but Google’s admission feels more candid than the usual corporate spin.

I’ve been watching this space long enough to remember when Google Cloud was the distant third, often losing deals to AWS and Azure on price or ecosystem. Now, the AI angle is changing the game. Companies like Anthropic, Character.ai, and a ton of startups are building on Google’s infrastructure, and the enterprise customers are following.

But capacity constraints are a double-edged sword. On one hand, it’s a good problem to have—demand is through the roof. On the other, it means Google is leaving money on the table, and in a competitive market, customers won’t wait forever. If Azure or AWS can provision faster, they’ll switch.

I’d be curious to see how Google plans to solve this. They’ve announced new data centers in Ohio, Indiana, and overseas, but those take time. The real bottleneck is chip supply. Google’s TPU v5 and v6 are custom silicon, which gives them an edge, but they still rely on TSMC for fabrication. And NVIDIA’s H100 and B200 GPUs are in even shorter supply.

For now, Google Cloud is riding the AI wave. But the capacity constraint is a reminder that even the biggest players have limits. The question is whether they can scale fast enough to keep the momentum going.

I’m not saying this is a crisis. Far from it. $20B in quarterly revenue is a milestone worth celebrating. But the fact that growth was “constrained” tells me the next few quarters are going to be interesting. Either Google ramps up capacity and accelerates, or competitors start eating into that demand.

Either way, it’s a good time to be in the cloud business. Just make sure you have enough servers.