The cloud numbers finally justify the spend
For three years, Google Cloud has been the hyperscaler perpetually described as catching up. That framing is now hard to sustain. Second quarter revenue reached $24.8 billion, up 82 percent year over year, and operating income nearly tripled to $8.8 billion from $2.8 billion a year earlier. Operating margin expanded from 20.7 percent to 35.6 percent in a single year, a pace of margin improvement that is rare for any business scaling this fast. CEO Sundar Pichai described the demand environment as strong across the board, and the unit economics back that up: this is no longer a division being subsidized to stay competitive with AWS and Azure.
The backlog tells the more interesting story for buyers. Google Cloud's remaining performance obligations hit $514 billion, up from $106 billion a year earlier, a nearly fivefold jump that reflects large committed contracts rather than pay-as-you-go usage. A meaningful share of that backlog is tied to TPU system sales rather than general-purpose compute, and Google has been explicit that the revenue ramp from those deals lands mostly in 2027, not this year. If you are negotiating a multiyear Google Cloud agreement now, the backlog number matters less than knowing whether your capacity is provisioned or merely promised.
A historic cash flow reversal
The cost of that growth showed up in a number Alphabet had never posted before: negative free cash flow. The company generated $39.1 billion in operating cash flow in the quarter, but capital expenditures of $44.9 billion outran it, leaving free cash flow at negative $5.9 billion. That compares with positive $10.1 billion in the first quarter of 2026. Alphabet has posted positive free cash flow every year since its 2004 IPO, through the 2008 financial crisis, the 2020 pandemic, and every prior AI spending cycle. This quarter broke that streak.
Alphabet's balance sheet can absorb it. The company holds $242.5 billion in cash and securities against $98.2 billion in long-term debt, giving it far more room than smaller cloud providers facing similar capex pressure. CFO Anat Ashkenazi framed the spending as return-driven rather than reflexive: "We are seeing very strong demand... our goal is to invest as long as we see an attractive return." That is the right discipline to demand from any vendor, and it is worth asking your own Google Cloud account team to show the utilization data behind it rather than accepting the growth narrative at face value.
Capex guidance keeps climbing
Alphabet raised full-year 2026 capex guidance to $195 billion to $205 billion, up from the $180 billion to $190 billion range set earlier this year. The quarter's spending broke down roughly 60 percent toward servers and AI accelerators and 40 percent toward data centers and networking, a split that shows the money is going into capacity that competitors and customers can eventually use, not just real estate. Ashkenazi did not soften the outlook for next year, telling investors that 2027 capex will increase significantly from already elevated 2026 levels.
That guidance raise triggered an immediate stock reaction: shares fell roughly 4 percent in after-hours trading, the market's clearest signal yet that investors are losing patience with capex increases unmoored from near-term revenue proof. Ashkenazi also flagged a constraint worth tracking closely if you are planning workload migrations: despite tripling compute capacity over three years, Google says demand is still outpacing supply investment, and component cost inflation is spreading beyond GPUs into the broader electronics supply chain.
What this means for enterprise buyers
Google Cloud's growth rate now exceeds both AWS and Azure on a percentage basis, but it is doing so from a smaller revenue base, which makes the comparison easier to hit and harder to sustain. Enterprise buyers evaluating a shift toward Google Cloud, particularly for AI workloads tied to TPU access, should treat the 82 percent growth figure as evidence of demand rather than evidence of available supply. The same capacity constraints reshaping AWS and Azure procurement conversations, longer lead times, committed-use pressure, and reduced negotiating leverage on spot pricing, are now squarely part of the Google Cloud conversation too.
The practical takeaway for CTOs and CIOs building multiyear cloud strategy is that capacity, not price, is becoming the primary negotiating variable across all three hyperscalers simultaneously. Google's backlog growth from $106 billion to $514 billion in a year means a large share of its future capacity is already spoken for by hyperscale AI customers. If your organization is counting on Google Cloud for TPU-backed workloads at meaningful scale, get commitments in writing now, because the vendors setting the terms on availability have shifted from sales teams to supply chains.
The bigger picture across hyperscalers
Alphabet's results land alongside a broader pattern: every major hyperscaler is now raising capex guidance mid-year while insisting demand still exceeds supply. That consistency across three independently reported earnings calls is itself a data point worth taking seriously rather than dismissing as vendor talk. It also means the industry-wide capacity crunch is a structural feature of the current AI infrastructure buildout, one that enterprise buyers need to plan around for at least the next 18 to 24 months rather than waiting out as a temporary planning miss.
For finance leaders watching capex-to-cash-flow ratios as a signal of vendor health, Alphabet's negative free cash flow quarter is a useful benchmark, not a red flag. The company still holds one of the strongest balance sheets in technology, and its willingness to run cash flow negative to secure AI infrastructure capacity suggests it views the current spending window as a use-it-or-lose-it moment for market share in enterprise AI compute. Vendors with weaker balance sheets making similar bets deserve more scrutiny than Alphabet does here.
