Alphabet Raises CapEx Guidance to $205 Billion as Cloud Backlog Hits $514 Billion and TPU Sales Begin Flowing Through P&L
Q2 2026 earnings call, July 22, 2026
Alphabet delivered its twelfth consecutive quarter of double-digit revenue growth, with consolidated revenue up 24% year-over-year to $119.8 billion, but the headline for investors is the magnitude of the capital spending ramp and the first real evidence of TPU hardware sales converting into cloud revenue. Management raised full-year 2026 capital expenditure guidance to a range of $195 billion to $205 billion, up sharply from the prior $180 billion to $190 billion outlook, and CFO Anat Ashkenazi reiterated that 2027 capex will step up again significantly, with more detail to come later. The increase, she said, is "primarily due to an acceleration in the delivery of capacity to meet growing demand," underscoring that Alphabet still views itself as supply constrained rather than demand constrained a year into the AI infrastructure buildout.
TPU System Sales Start Hitting Revenue, With Margin Give-and-Take
The most concrete new data point this quarter is that Alphabet began recognizing revenue from third-party TPU system sales for the first time, after delivering systems into customer data centers. Ashkenazi was explicit that this is still early: "This quarter was a small amount of the total... that total agreement will continue to ramp up throughout 2026, but then you'll see the vast majority of the revenue from that agreement comes through in 2027." Importantly, she noted Cloud revenue growth "accelerated meaningfully even after excluding the impact of TPU system sales," suggesting the core GCP franchise is not simply being flattered by one-time hardware bookings. TPU agreements are embedded within the $514 billion cloud backlog, which grew by more than $50 billion sequentially, though the majority of that backlog remains standard GCP service contracts, with just over 50% expected to convert to revenue over the next 24 months.
On margins, Ashkenazi declined to break out TPU-specific profitability but framed the hardware business as an expansion of total addressable market rather than a dilutive distraction. She flagged two near-term margin headwinds investors should model: increased use of third-party compute capacity as a "bridging strategy" while Alphabet builds out internal capacity, which will create "modest margin pressure" in the third quarter, and continued integration costs from the Wiz acquisition. Despite these puts and takes, Cloud operating margin still expanded dramatically, from 20.7% a year ago to 35.6% this quarter, with operating income more than tripling to $8.8 billion on revenue growth of 82% to $24.8 billion.
Gemini 4 Pretraining Underway, With Pichai Conceding Coding Gap
CEO Sundar Pichai disclosed that Alphabet has "started our most ambitious pretraining run yet for Gemini 4," a notable escalation in scale and ambition following the Gemini 3.5 Pro testing phase currently underway. More strikingly, Pichai directly acknowledged competitive weakness in one area: "There are areas where we have acknowledged we need to improve, coding and agentic coding is an example of that." He pointed to Gemini 3.6 Flash's 10-point-plus jump on the DeepSWE coding benchmark versus 3.5 Flash, achieved in just six weeks, as evidence of rapid iteration, and said Alphabet is targeting a near-monthly release cadence as it builds toward Gemini 4. This is a rare instance of Alphabet management explicitly naming a capability gap versus rivals rather than simply asserting frontier parity, and it's a data point worth tracking given how central coding and agentic workflows have become to enterprise AI monetization narratives across the sector.
Compute Allocation Priorities: Frontier Development Comes First
Multiple analysts pressed on how Alphabet rations scarce compute, and Pichai's answer provides a useful hierarchy for investors modeling internal versus external capacity trade-offs. The baseline allocation goes to frontier AGI model development, followed by core consumer products including Search and YouTube, then Cloud serving for Vertex AI, Gemini Enterprise, and first-party model inference. External TPU placements, including the previously disclosed Blackstone data center project, serve as a release valve to monetize demand that cannot be absorbed internally. On the bridging capacity strategy, Pichai characterized it as economically rational despite near-term margin drag: "Are you willing to take upfront a six-month deal to be able to serve that customer in what is a multiyear opportunity where the margins and the returns are very, very attractive over that multiyear horizon."
Search Monetization Efficiency Improving Even as AI Costs Rise
Search and Other revenue grew 17% to $63.3 billion, with Philipp Schindler attributing gains to deeper Gemini integration across ad quality, advertiser tools, and query understanding. A specific proof point: shopping ads saw a 20% improvement in relevant ad matching. AI Max, now out of beta with 500,000 advertisers, is delivering an average 15% lift in conversions or value at similar return on ad spend, with AAA Auto Club Enterprises citing a 17% conversion improvement and an 11% reduction in cost per lead. Pichai also noted that engineering optimizations pushed the cost of serving AI Mode responses to its lowest level since launch, even as capability has increased, a signal that unit economics on the consumer AI product are improving rather than deteriorating, which should ease some investor concern about margin dilution from AI Overviews and AI Mode rollout. AI Mode has now surpassed 1 billion monthly active users since its global expansion last October.
Capital Structure Shift Toward Debt and Equity Funding
Ashkenazi provided new detail on how Alphabet is financing the buildout beyond operating cash flow. The company's debt portfolio has expanded from roughly $16 billion a year ago to approximately $100 billion across multiple currencies, and Alphabet recently tapped equity markets as well, though Ashkenazi indicated no further equity raises are planned outside of an at-the-market program specifically earmarked to cover taxes on stock-based compensation. This is a meaningful shift in funding posture for a company that has historically self-funded through operating cash flow, and it reflects the sheer scale of the capex ramp: free cash flow was negative $5.9 billion in the quarter, though it remains positive at $53.3 billion on a trailing-twelve-month basis.
YouTube and Advertising Strength Tied to World Cup, but Broad-Based
YouTube ad revenue grew 13% to $11.1 billion, aided by 1.7 billion unique viewers of World Cup content, making it the most-viewed World Cup in YouTube's history. Network advertising revenue, by contrast, declined 1%, continuing a trend of secular weakness in that legacy segment. Growth in Search was described as broad-based across retail, finance, tech, and media verticals rather than concentrated in any single category, and Schindler pointed to new commerce infrastructure, including the Universal Commerce Protocol adopted by Target and Steve Madden, and a new Universal Card enabling cross-retailer checkout, as incremental monetization surfaces still in early innings.
Waymo Structure Question Deferred
Asked directly whether Waymo could eventually be spun out or restructured given its scaling momentum, Pichai declined to engage with the hypothetical, stating simply that Alphabet is "really focused on executing and scaling up Waymo" under its current bet structure. The company did disclose that Waymo's newest vehicle, Ojai, powered by a sixth-generation Waymo Driver system, has begun carrying public riders, and Wing has now completed over 1 million home deliveries. Other Bets remains a modest drag on consolidated results, with $382 million in revenue against a $1.8 billion operating loss for the quarter.