Google Cloud hits $20B, but the real story is what held it back

Google Cloud just hit a milestone that would have seemed absurd a few years ago: $20 billion in a single quarter. That’s real money, and it’s largely thanks to the AI boom that’s been reshaping cloud infrastructure spend across the industry.

But here’s the part that caught my attention. Alphabet’s CFO Ruth Porat didn’t just celebrate the number. She explicitly said growth was “capacity-constrained” during the quarter. Translation: they could have sold more, but they simply couldn’t spin up enough compute fast enough.

This is a recurring theme across the big three cloud providers right now. Everyone is scrambling to build out AI-optimized data centers, and GPUs are still hard to come by at scale. Google’s situation is particularly interesting because they’ve been investing in custom TPUs for years, yet even that vertical integration isn’t enough to keep up with demand.

The $20.3 billion figure represents 28% year-over-year growth, which is impressive by any standard. But when you hear “capacity-constrained,” you start wondering what that number could have been with unlimited hardware. 30%? 35%? More?

What I find telling is that Google didn’t try to spin this as a positive. No “we’re managing demand carefully” or “we’re prioritizing quality over quantity.” They just acknowledged the bottleneck. That’s refreshingly honest for an earnings call, even if it also signals to investors that there’s untapped revenue sitting on the table.

For context, AWS and Azure are facing similar constraints, though they tend to frame it differently. AWS talks about “long lead times” for certain instances. Azure mentions “supply chain dynamics.” Google’s directness feels like a subtle flex: we have the demand, we just need the hardware.

The real question is how long this constraint lasts. Google is pouring billions into new data centers across the US, Europe, and Asia. But building these facilities takes time, and even once they’re operational, filling them with accelerators isn’t instant. I’d expect capacity to remain a talking point for at least the next two to three quarters.

One thing that’s easy to overlook: this $20B quarter isn’t just about AI inference or training workloads. Google Cloud’s broader portfolio, including Workspace, data analytics, and security, is also growing. AI is the headline, but the underlying platform strength matters too.

Still, the capacity constraint is the real story here. It tells you that enterprise AI adoption isn’t slowing down. It tells you that Google’s custom silicon strategy, while smart, hasn’t completely insulated them from the global hardware crunch. And it tells you that the cloud wars are far from settled.

I’ll be watching next quarter’s numbers closely. If Google can resolve some of these capacity issues, that $20B could look like a floor, not a ceiling.