Alphabet said on 28 October 2026 that Google Cloud revenues increased 103% to $30.7 billion in the third quarter, and that “TPU systems sales were also a meaningful contributor to revenuegrowth.”
On the earnings call the same day, finance chief Anat Ashkenazi said: “GCP accounted for the vast majority of the revenue generated in Q3.” She also said Google had started renting outside capacity: “given the supply-constrained environment, we started utilizing third-party compute capacity at the end of Q3 and will ramp usage in Q4.”
So Google is selling its own AI chip systems to other companies, including some it competes with, while renting compute from third parties to serve its own cloud. Several analysts on the call asked about that trade-off. This piece sets out what the executives said; call quotations come from a transcript of the webcast.
Business Pill · Vertical Integration
An evergreen Business Engineer explainer on vertical integration, why every layer takes a cut. It explains the general idea only and does not discuss TPU systems.
The key insight: By our arithmetic from Alphabet’s release, the $3.2 billion of cash Alphabet put into inventory in the third quarter equals about 36% of its negative $9.1 billion free cash flow, while its finance chief described Google Cloud as supply-constrained.
Thomas Kurian, chief executive of Google Cloud, said in his prepared remarks: “In addition to our work with Anthropic, both directly and via Broadcom, and public reference customers like Apple, who are scaling pre-training of their foundation models on the latest generation of TPUs.” He said “nearly hundred customers use TPUs today”. Tonight’s TPU customers piece lists the other names he gave.
Answering Goldman Sachs analyst Eric Sheridan, Kurian described the path: “we have over a hundred significant TPU customers”, who “start in our cloud. They learn to use TPUs.” He said Google was now “making TPU systems available in their data centers. We’re also broadening availability to other providers.”
He added: “we announced a partnership with Oracle, and we’ve launched, along with Blackstone, an agreement for Crux AI, a joint venture to offer customers TPU systems.”
Two uses for the same TPU systems, as Alphabet’s executives described them on the 28 October 2026 earnings call: running Google’s own supply-constrained cloud, or selling systems to customers and partners. The inventory share of free cash flow is our arithmetic. Sources: Alphabet Q3 2026 earnings release; transcript of the Q3 2026 earnings call webcast.
How Google Says It Splits Scarce Compute
Sheridan asked Sundar Pichai how Google sets priorities for compute as it becomes available. Pichai said compute allocation is an area where Google spends a lot of time because of the forward planning needed, and that each business submits long-range plans under an ROIC framework.
He said: “I would broadly think of all of this as a long-range planning process, not a short-term allocation exercise.”
Asked by AllianceBernstein’s Mark Shmulik how much capacity had held back cloud growth, Kurian said: “we could obviously serve a lot more demand with capacity, but we are managing all of these parameters and growth carefully.”
Ashkenazi said the third-party capacity “came online only at the very end of Q3 and therefore had limited negative impact to our margin in the quarter.” Looking ahead, she said: “We expect margin headwinds in the near term as we utilize this capacity.”
Why Sell to Competitors
MoffettNathanson’s Michael Nathanson said the call had shown “a willingness to work with some of your competitors or companies that we may view as competitors in models and cloud”, and asked why that is an opportunity rather than a risk.
Pichai said Google builds some things as products and others as platforms, and “you need to have a clear view on which is which.” He said Google works “on our models with other companies, including Apple, with whom we compete in many, many areas.”
“Similarly, we view our TPU systems as an important platform we are building,” he said. Answering Shmulik earlier in the session, he said: “in the semiconductor industry, economies of scale matter.”
What Selling Does to Cash
Barclays’ Ross Sandler asked what it means for Alphabet’s capex as it starts to “hand off some of the future CapEx to the AI labs”. Ashkenazi said a signed agreement shows up first in backlog, and then Google starts building inventory.
“So it’s not the traditional kind of CapEx as you’re thinking about this,” she said, “but rather a different cycle of cash flow generation.” Kurian said Google procures components for that inventory only after contracts are signed, and that placing systems in customer and third-party facilities “offsets the need for us to have to pre-build facilities to host these” systems.
Alphabet’s release shows inventory of $13.3 billion at 30 September, against $2.4 billion at the end of 2025, and an inventorycash outflow of $3.2 billion in the quarter. Free cash flow was negative $9.1 billion. By our arithmetic, the quarter’s inventory build equals about 36% of that shortfall. The inventory piece and the commitments piece cover the balance-sheet side.
The Margin Question
Bank of America’s Justin Post asked why selling systems is “a good way to use all this capacity”. Kurian said: “we don’t break out margins for individual products or infrastructure components.”
He said: “Given the price-performance benefits of TPU systems relative to competitors, growing demand for accelerated compute and the scarcity of accelerated compute, we’re very focused on continuing to improve margins as we scale the business.”
The Structural Read
As we read the call, the same TPU systems now have two uses: running Google’s own cloud, which Ashkenazi called supply-constrained, and selling them to customers that include companies Google competes with.
Pichai framed the sales as a platform choice: “we view our TPU systems as an important platform we are building.” Kurian tied them to margins: “we’re very focused on continuing to improve margins as we scale the business.”
The Business Engineer premium analysisThe TPU Trap examines this trade-off from the Q2 print; its argument is the analysis’s own, not a claim of this piece.
FOR GOOGLE CLOUD Ashkenazi said Google started using third-party compute at the end of Q3, will ramp it in Q4, and expects near-term margin headwinds.
FOR ALPHABET’S CASH FLOW Ashkenazi said a signed system sale shows up first in backlog and then in inventory, with cash coming in as Google delivers and bills.
FOR TPU BUYERS Kurian said Google is making TPU systems available in customers’ data centers and through other providers, including Oracle and Crux AI.
The framework argues: “The winning strategy in every major platform shift has followed a consistent pattern: Close the SCARCE layer (keep proprietary) + Open the ABUNDANT layer (commoditize)”.
Kurian described accelerated compute as scarce on the call, and Google is opening its TPU systems to outside buyers. The framework does not discuss TPU systems; how the two fit is the question the call left open.
What Is Not Established
Alphabet does not report TPU system revenue separately from the rest of Google Cloud, so the share of the quarter’s 103% growth that came from system sales is not known. Ashkenazi said Cloud revenuegrowth excluding TPU systems “was again incredibly strong” but gave no figure.
The executives did not say how much capacity goes to Google’s own products, to Cloud customers and to system sales, or how much third-party compute Google is renting.
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
Alphabet’s executives described two uses for the same TPU systems: running Google’s own cloud, which its finance chief called supply-constrained, and selling them to customers including Anthropic and Apple. They framed the sales as a platform business that builds scale, avoids pre-building some facilities and turns into cash after delivery, while the inventory behind them is already on the balance sheet.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. Call quotations come from a transcript of the 28 October 2026 earnings call webcast; revenue, inventory and cash-flow figures come from Alphabet’s third-quarter 2026 results release. The inventory share of free cash flow is our arithmetic; linked earlier pieces carry their own sources.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
