Earnings Recap — Q2 FY2026
CY Q3 2026 · Reported July 22, 2026 · Beat 7 of last 7 quarters
The Buildout is a website built on agentic AI tools, currently in beta, and could have factual errors.
Alphabet's massive CapEx raise and supply-constrained commentary underscore the accelerating AI infrastructure buildout. The company's move to sell TPU systems directly to customers and use third-party capacity as a bridge highlights the industry-wide scramble for compute. Cloud backlog growth to $514B signals sustained enterprise AI demand, reinforcing the need for continued data center and chip investment.
Alphabet delivered 24% YoY revenue growth to $119.8B, with Search up 17% and YouTube ads up 13%. Google Cloud revenue grew 82% to $24.8B, with backlog reaching $514B, and Cloud operating margin expanded to 35.6%. The company began recognizing revenue from TPU system sales in Q2. CapEx was $44.9B, leading to negative free cash flow of $5.9B. Net income was boosted by $98B in other income, primarily unrealized gains on equity securities.
Management raised full-year 2026 CapEx guidance to $195B–$205B (from $180B–$190B) and reiterated that CapEx will increase significantly in 2027. They expect to use third-party capacity in Q3 as a bridging strategy, which will create modest near-term margin pressure, while TPU system sales will ramp through 2026 with the vast majority of revenue recognized in 2027. Cloud margins face near-term headwinds from third-party capacity and the Wiz integration, but the long-term demand environment is described as 'healthier' than a year ago. Management emphasized a disciplined ROIC framework and a full-stack approach to capture the AI opportunity, with continued hiring in AI and Cloud and marketing investments to support AI products. They also noted a slight FX headwind to Q3 consolidated revenue and lapping of Search acceleration from Q3 2025.
“We are seeing strong diversified demand across products, customers, geographies and industries.”
on Cloud demand
“We are still in a supply-constrained environment. I think we've said this now for multiple quarters in a row. And we are seeing very strong demand, both from external cloud customers as well as across the business.”
on Supply constraints
“We are both very committed and very confident of being at the frontier. For the next generation of frontier, you're going to need a much larger base models. We are now training Gemini 4, and we're being very ambitious with it.”
on Model frontier
How has your view on the size and timing of GenAI ROIC changed versus a year ago? And how is your budgeting philosophy for forward CapEx changing given supply constraints?
Sundar said they are 'more bullish' on the opportunity, citing early innings across consumer, enterprise, and developer adoption. Anat said they invest as long as returns are attractive, taking a multiyear view, and that demand still outpaces investment despite significant capacity increases.
How confident are you that Gemini models can remain at the frontier, and how do you plan to close the gap in coding? Also, how do you think about optimal capital structure after recent equity and debt raises?
Sundar acknowledged coding as an area to improve but highlighted progress with 3.6 Flash and the ambitious Gemini 4 training run. Anat explained the capital structure approach: cash from operations first, then debt, and the equity raise was to maintain a resilient balance sheet; no plans to return to equity markets except for the ATM related to SBC taxes.
What are your key learnings scaling TPUs, and how do you balance external vs internal demand? Also, can you give more granularity on TPU revenue recognition and margin impact?
Sundar said TPUs are first allocated to frontier model development, then to serving consumer and enterprise products, with external sales balanced via data center placements like the Blackstone project. Anat said TPU system sales are in the cloud backlog, with revenue recognized upon delivery; Q2 was small, ramping through 2026, with the vast majority in 2027.