Google Cloud just posted its first $20 billion quarter, and the bigger story is what it couldn’t deliver. According to TechCrunch AI, Alphabet’s cloud arm grew revenue 63% year-over-year in Q1 2026, but CEO Sundar Pichai warned analysts on Wednesday’s earnings call that the business is “compute constrained” and that revenue would have been higher if Google could meet demand.
The backlog tells the story. Google Cloud’s order book doubled in a single quarter to $462 billion, and the company expects to work through only half of it over the next 24 months. Pichai framed the constraint as proof of momentum, not a stumbling block.
What’s driving the surge
AI workloads were the biggest single contributor. TechCrunch AI reports that products built on Google’s generative AI models grew nearly 800% year-over-year. Other standout numbers from the call:
- Gemini Enterprise revenue grew 40% quarter-over-quarter
- API traffic hit 16 billion tokens per minute, up from 10 billion in Q4
- New customer acquisition doubled year-over-year
- Deals between $100 million and $1 billion doubled, with multiple “billion-dollar-plus” contracts signed
- Existing customers outpaced their initial commitments by 45% quarter-over-quarter
Google Cloud Platform itself grew faster than the broader Cloud division, which also includes Workspace, data analytics, and infrastructure services.
Why the constraint matters
This is the second hyperscaler in as many weeks to admit it can’t build capacity fast enough. AWS just posted its strongest growth in 15 quarters and flagged similar supply pressure. The pattern is clear: enterprise AI demand has front-run the physical infrastructure needed to serve it, and the bottleneck is now data centers, power, and custom silicon.
For Google specifically, the constraint sits at the TPU layer. Pichai pointed to “strong demand” for both TPU hardware and data center capacity, and noted that Google now sells TPUs directly to some customers in addition to renting compute through the cloud. That direct-hardware play is a relatively new revenue lane, and it’s expanding alongside infrastructure rentals.
What stands out here is the size of the backlog relative to current revenue. A $462 billion order book against a $20 billion quarter implies roughly five and a half years of locked-in demand at current run rates, even before accounting for new deals. That’s an extraordinary pipeline, and it explains why Pichai keeps tying capital allocation decisions to return on invested capital rather than chasing every workload.
What practitioners should expect
For enterprise teams building on Google Cloud, three near-term implications:
- Capacity allocation will get tighter. Expect longer lead times for new TPU reservations and pricier spot pricing for Gemini API workloads.
- Big contracts get priority. Pichai’s emphasis on nine-figure and ten-figure deals signals that Google is sorting customers by commitment size. Smaller teams should plan multi-cloud fallbacks.
- Hardware buys may make sense. With direct TPU sales now a real channel, large AI shops have an option to bring inference in-house rather than wait for cloud capacity.
The capital expenditure cycle isn’t slowing down. Alphabet, Microsoft, Amazon, and Meta are all pouring tens of billions into compute infrastructure, and the demand signal from this quarter suggests the spending will keep climbing through 2026 and beyond.
Full breakdown of Google’s Q1 numbers is at TechCrunch AI.