Google Cloud's 82% Surge and $514 Billion Backlog: What It Means for Buyers

Google Cloud revenue grew 82% in the second quarter to $24.77 billion, Alphabet's contracted cloud backlog reached $514 billion, and the company raised its 2026 capital spending forecast to as much as $205 billion. Investors sent the shares down more than 6% the next day. For enterprise technology leaders, the July 22 results say less about Alphabet's stock and more about the market they are buying in: demand for AI cloud capacity is running ahead of supply, and that shifts negotiating power toward the providers.

What did Alphabet report?

Alphabet's revenue rose 24% to $119.8 billion, and Google Cloud revenue climbed from $13.62 billion a year earlier, 9to5Google reported. Alphabet said nearly 90% of the Fortune 100 now use Gemini Enterprise and that its models process 22 billion API tokens per minute.

The cloud backlog, or remaining performance obligations, came in at $514 billion against analyst projections of $488.1 billion, Yahoo Finance reported. Alphabet raised its 2026 capex outlook to $195 billion to $205 billion, from $180 billion to $190 billion, above the $186.4 billion analysts expected, and said spending will grow significantly again in 2027.

Second-quarter capex alone was $44.9 billion, and cloud growth accelerated from 48% in the fourth quarter of 2025, AI Weekly noted. Management said just over half of the backlog should turn into revenue over the next 24 months, and that most revenue from existing TPU system sales agreements will be recognized in 2027, according to a summary of the earnings call.

Why did the stock fall after a beat?

The bear case is about cash, not demand. Free cash flow was negative by about $5.9 billion in the quarter, AI Weekly reported, citing a Seeking Alpha summary. Headline net income was inflated by about $98 billion of unrealized gains on equity stakes, including Anthropic and SpaceX, which says little about operating performance.

Yahoo Finance also cited a Bloomberg report that Google delayed its Gemini 3.5 Pro model over capability concerns. A Google spokesperson said the company is testing 3.5 Pro and other models with partners. For investors, a delayed flagship model plus a larger spending plan raised questions about return on capital.

The bull case is that the backlog makes the spending less speculative. As DIGITIMES framed it, the debate has moved from whether Alphabet can afford its AI build-out to whether it can build fast enough. A $514 billion order book, with roughly half due within two years, is contracted demand that capex is chasing.

Both views can be true. Alphabet is spending ahead of cash generation because customers have signed for capacity it has not yet built.

What does a capacity-constrained cloud mean for enterprise buyers?

Big technology companies have said they are capacity-constrained, Yahoo Finance noted, and Google Cloud's growth rate is consistent with that. Three implications follow for CIOs and infrastructure leaders.

Committed capacity beats spot capacity. When providers have a half-trillion-dollar backlog, the customers who signed multi-year commitments get served first. Teams planning large inference or agent workloads for 2027 should expect to reserve capacity rather than assume it will be available on demand.

Discounts get harder to win. Strong demand reduces a provider's need to cut price on AI services. Enterprises with flexible workloads can still negotiate, but the leverage that came from a buyer's market in general-purpose cloud does not carry over to scarce accelerators.

Hardware options are widening. Alphabet's comment that TPU system sales revenue lands mostly in 2027 shows Google selling its chips into customer data centers, not only renting them in its cloud. For the largest buyers, that adds an on-premises or hosted path to the usual menu.

How should enterprise leaders respond?

The practical answer is cost discipline and architectural flexibility, the same themes AI Connect speakers raise when AI moves from pilot to production. In a AI Connect panel on AI infrastructure at scale, AtScale's Dave Mariani argued that infrastructure must be open because "we don't know where this stuff is going." Locking every workload into one provider's accelerators and tooling is a bet on that provider's capacity plans.

Cost overruns are the other risk. SandboxAQ's Stefan Leichenauer warned in a separate AI Connect panel that teams can "rack up huge costs accidentally" with cloud AI services, and urged a clear owner for those decisions, a point covered in our piece on future-proofing AI infrastructure.

For enterprise leaders, the takeaways are concrete:

• Forecast AI compute needs for 2027 now and decide which workloads justify reserved capacity.
• Keep at least two model and infrastructure options production-ready for critical use cases.
• Tie every large commitment to a measured business outcome, so spending can be defended if prices rise. AI Connect's coverage of why AI pilots fail to scale lays out a practical test for deciding what deserves that investment.

Alphabet's quarter confirms that enterprise AI demand is real and contracted. It also confirms that the supply side will set the pace, and enterprises that plan for scarcity will have more options than those that assume abundance.

Frequently asked questions

How fast did Google Cloud grow in Q2 2026?

Google Cloud revenue rose 82% year over year to $24.77 billion in the quarter, up from $13.62 billion a year earlier.

What is Google Cloud's backlog?

Alphabet reported $514 billion in remaining performance obligations, and management said just over half should be recognized as revenue over the next 24 months.

Why did Alphabet stock fall after beating estimates?

Shares fell more than 6% after Alphabet raised its 2026 capex forecast to $195 billion to $205 billion, amid concerns about cash flow and a reported delay to Gemini 3.5 Pro.

Filed under: Data & AI Infrastructure, AI Strategy & ROI