Sundar Pichai spent Alphabet’s Wednesday earnings call doing something he is not used to doing: arguing that Google is not, in fact, falling behind in artificial intelligence.

Analysts wanted to know why the next Gemini model had slipped, and whether the company had quietly ceded the frontier to OpenAI and Anthropic.

“We’ve had clearly frontier models,” Pichai said on the call. “There are many attributes on which we are still at the frontier; there are areas where we’ve acknowledged we need to improve and coding and agentic coding is an example of that.”

For a company that likes to remind everyone it invented the transformer, the admission was unusually candid.

The remarks answered a run of pointed questions from JPMorgan’s Doug Anmuth and Barclays’ Ross Sandler, who pressed him on a delayed Gemini Pro and a release schedule that has trailed rivals.

Google’s next Pro model is months behind schedule, and coding is exactly the ground on which OpenAI’s Codex and Anthropic’s Claude have been peeling off developers.

Pichai’s answer was to promise more, and sooner. He said Google had begun its most ambitious pre-training run yet for Gemini 4, told analysts “I think people will be pleased” when it lands, and signalled that future models would arrive at close to a monthly cadence.

He also leaned on demand as evidence that the doubters had it wrong. Google remains “supply constrained,” Pichai said, which he cast as a sign of momentum rather than strain.

Gemini models now process 22 billion API tokens a minute, and nearly 500 cloud customers have each run more than a trillion tokens over the past year.

The figures underneath the argument were strong. Alphabet posted revenue of $119.8 billion for the quarter, up 24% on a year earlier, while Google Cloud grew 82% to $24.8 billion, comfortably ahead of the roughly 64% analysts had pencilled in. Cloud backlog reached $514 billion.

Advertising, still the engine, did its share alongside cloud. YouTube brought in $11.06 billion in ad revenue, up from $9.79 billion a year earlier, and the wider advertising business kept growing even as search shifted towards AI-generated answers.

Consumer numbers held up too. The Gemini app hit 950 million monthly active users, its daily total tripling over the past year, and nearly 90% of the Fortune 100 now run Gemini Enterprise.

Google’s full-stack bet, from custom chips to models to agents, is the version of events Pichai would like investors to read.

Net income came in at a startling $112.1 billion, though most of that was an accounting mirage. Around $80 billion of it was an unrealised gain on Alphabet’s stake of roughly 14% in Anthropic, whose valuation tripled during the quarter.

What unsettled the market was spending. Alphabet lifted its 2026 capital expenditure guidance by about $15 billion, to between $195 billion and $205 billion, extending a year of record AI capex across Big Tech, and management signalled the bill would climb again in 2027.

The message to investors was that the constraint is chips and data centres, not demand, and that Google intends to keep buying its way to the front.

Wall Street was not fully persuaded. Shares fell more than 3% after hours, and are down roughly 9% since late April on the Gemini delays and a string of senior departures.

The anxiety is not new. The next Gemini Pro was reportedly expected in June, and its slip has coincided with a run of departures from Google’s AI ranks, feeding a story in which the company that published the transformer paper keeps struggling to turn research into shipped product.

The concession on coding was the tell. A year ago Google was beating estimates and closing on Nvidia as the most valuable company in the world; now its chief executive is answering questions about whether it can keep pace. The next Gemini, whenever it ships, will have to make his case for him.