Oct 28, 2026 earnings call — Gemini 4 models and Cloud TPU scaling
Alphabet revenues grew 25% year-over-year and Google Cloud revenues grew 103% in the quarter.
Revenue $127.6B (est $128.0B) · EPS $2.12 (est $3.02)
Key quotes
"Our first-party model APIs are now processing nearly 32 billion tokens per minute, 40% more than last quarter and more than 4x what we processed a year ago."
"TPU 8t and 8i offer up to 2.7x better training performance and 80% better inference price performance, respectively, versus our previous generation of TPUs."
"Based on these collective results, we're now scaling high-quality ads within AI Overviews and AI Mode to more users and queries."
Analyst Q&A 7
AI productivity and Search ad impact
Q (Douglas Anmuth · JPMorgan): Sundar: how AI models create a "golden age" of productivity. Philipp: why query growth is fast but Search revenue growth slowed—are AI Overviews and AI Mode showing fewer ads?
A: Sundar said new internal models (3.8 Flash, Argon, Carbon) are driving big productivity gains for engineering and product innovation, and Carbon deployment just started so he expects more rollout and innovation. Philipp said Search grew 12% year-over-year to $63 billion, cited FX and vertical-specific factors, and noted AI Mode had lower monetization but they are scaling ads in AI Overviews and AI Mode.
"Yes, Doug, on the second part of your question, look, in Q3, Search grew 12% year-over-year to $63 billion." — Philipp Schindler
Scaling agentic products and FX impact on Search
Q (Brian Nowak · Morgan Stanley): What technical hurdles remain to ship agentic products to all users (efficiency, capacity, readiness)? And how should we think about the Search comparison factor into Q4 and early 2026?
A: Sundar said they are making distilled Carbon models to deploy to 1 billion users efficiently and that they need a robust harness, secure runtime (GCP), and connectors; enterprise rollout came first and consumer rollout will follow later this year. Anat said FX spot rates are a sequential headwind into Q4 and that they are lapping about 3 points of growth from late 2025, which will continue to affect comparisons.
"So where we are today with current spot rates, we expect that this will be somewhat of a headwind to Q4 results." — Anat Ashkenazi
How compute is allocated across Google and Cloud
Q (Eric Sheridan · Goldman Sachs): How does Alphabet prioritize compute allocation across businesses as capacity becomes available? And within Cloud, how do you view applying compute to customer needs versus external TPU systems?
A: Sundar said compute allocation is part of long-range planning using disciplined ROIC frameworks and that AI models are foundational across products. Thomas said Cloud serves a broad set of customers, many start using TPU systems in Google Cloud, and they are building silicon (8t, 8i), optimizing software (PyTorch), and broadening distribution via partners and customer data center deployments.
"We have over 100 significant TPU customers who start in our cloud." — Thomas Kurian
How TPU system sales show up financially
Q (Ross Sandler · Barclays): When you resell TPU systems, what percent of system cost shows up as revenue and how does that affect Cloud margins and Alphabet CapEx intensity as labs bring their own infrastructure?
A: Anat said they do not provide product-level financials but explained TPU systems are full solutions (hardware, storage, networking, software) and sales show up first in backlog while they build inventory, so cash goes out to build inventory and then cash comes in when they deliver and bill. She said some system sales are recorded already and more revenue from system sales should come through 2027.
"When we signed an agreement with the customer to sell TPU systems, once we sign that agreement, you'll first see that showing up in our backlog and then we start building inventory." — Anat Ashkenazi
Will agentic change Search and TPU margins
Q (Justin Post · BofA): Could agentic workflows hurt Search or increase monetization long term? And are TPU systems a good, profitable business?
A: Philipp said agentic experiences are additive and they are building commerce blocks (cart, agents, payments) to automate grunt work while preserving discovery and choice, which creates new advertising models and monetization opportunities. Sundar called agents an expansion opportunity that will add use cases; Thomas said they don't break out margins but see price-performance benefits, committed inventory tied to signed contracts, and that placing TPU systems externally broadens the market and offsets building hosting facilities.
"we view agentic experiences as clearly additive." — Philipp Schindler
Early agentic use and Cloud growth excluding TPUs
Q (Mark Shmulik · AllianceBernstein): Have early agentic use cases changed your view, especially on consumer productivity? And how fast is Cloud growth excluding TPU system sales and was Cloud growth constrained by capacity?
A: Sundar said they see agentic task use in the Gemini app and that long tasks must cross a quality threshold to drive repeat use; he expects strong progress later this year and in 2027. Thomas said Google Cloud had record growth in Q3 with Cloud revenues up 103% to $30.7 billion, GCP drove most growth, Cloud growth excluding TPU system sales was also very strong, and demand is broad though they manage capacity and growth carefully.
"Google Cloud segment delivered record growth in the third quarter with Cloud revenues up 103% to $30.7 billion." — Thomas Kurian
Working with competitors and agentic safety/trust
Q (Michael Nathanson · MoffettNathanson): Why work with companies that compete with you in models and cloud, and what have you learned about consumer concerns on safety and trust for agentic tools?
A: Sundar said they distinguish products vs platforms and when building platforms they make them work for everyone; TPU systems are a platform they want to make world-class and test against market forces. On safety, he said people are clearly concerned about safety and trust, and Google relies on decades of building safe experiences, extensive testing, and real-time monitoring as key differentiators to protect trust and privacy.
"I mean, to the earlier question, and I think I should have emphasized, people really -- when people are approaching these agentic experiences, we clearly see people are concerned about safety and trust." — Sundar Pichai