Google's cloud doubles, Azure jumps 48%, and OpenAI reportedly shops for Google chips

Wednesday, October 28, was report card day for two of the biggest companies in tech. Alphabet and Microsoft both published their results for the July to September quarter, and both told a similar story: businesses want a lot more AI computing power, and the companies that rent it out are growing fast. What investors made of it was a different matter, and the two stocks went in opposite directions.

Around those results, the rest of the day filled in the picture. A report said OpenAI plans to buy Google's own AI chips, Google opened its Gemini assistant and Search to more outside apps, Anthropic announced a large discount on Claude for US governments, and President Trump said he wants to ban the term AI in government documents.

Google's cloud business doubles in a year

Big public companies report their numbers every three months, and investors read them for clues about where things are heading. Alphabet, Google's parent company, reported revenue of $127.6 billion for the quarter, up 25% from a year earlier.

The standout was Google Cloud. That's the part of Google that rents out computing power, storage and AI tools to other businesses, so they don't have to build their own data centers. Its revenue rose 103% to $30.7 billion, up from $15.2 billion a year earlier. In Alphabet's release, CEO Sundar Pichai said the growth was "powered by huge demand for our AI infrastructure and AI solutions." He also pointed to strong demand for Google's TPU systems, the company's own AI chips, and the release says sales of those systems were a meaningful contributor to cloud growth. That detail matters for one of the day's other stories, below.

Google also gave a few numbers about how much its AI is being used. By its own count, the Gemini app has passed one billion users, and Alphabet calls it the fastest-growing product in its history. The company says its AI models, offered to developers through its own services, now process nearly 32 billion tokens a minute. A token is a small chunk of text, roughly part of a word, and it's the unit AI companies use to measure, and charge for, the work their models do.

All of this costs money. Alphabet spent $51.4 billion on property and equipment in the quarter, which mostly means data centers, servers and chips. A year earlier, that figure was $24.0 billion, so spending has more than doubled.

So why did Alphabet's shares fall about 8% after the results, according to TechCrunch? Search. Revenue from Google Search and related products grew 12% from a year earlier, to $63.4 billion. That is still growth, but it was 17% the quarter before and 19% in the first quarter of the year. Search is still Google's biggest business, so investors watch that number closely.

Google's answer is to put ads into its AI answers. AI Mode is the chatbot-style version of Google Search, where you ask a question in plain language and get a written answer instead of a list of links. Google is now bringing ads in AI Mode to markets outside the US for the first time. On the call with investors, Google's chief business officer Philipp Schindler said the company is "now scaling high-quality ads within AI overviews in AI mode." Google also told investors that people's AI Mode questions are three times longer than traditional searches on average, and that one in five AI Mode searches involves voice, images or documents.

Microsoft's Azure grows 48%

Microsoft's results got a warmer reception. According to SiliconANGLE, revenue for the quarter was $92.28 billion, up 19%, ahead of what analysts had expected. Microsoft's shares rose about 4% in trading after the market closed.

The headline number was Azure, Microsoft's cloud platform and its equivalent of Google Cloud. Azure grew more than 48%, and for the first time Microsoft said exactly how much money it brings in: $33.17 billion in the quarter. SiliconANGLE notes the figure is part of a new way Microsoft reports its results.

Microsoft also said the number of people using Microsoft 365 Copilot, the AI assistant built into apps like Word and Excel, rose 50% compared with the previous quarter. CEO Satya Nadella did not say how many of those users are paying for it.

The most telling line came from Amy Hood, Microsoft's chief financial officer. "We still have demand that exceeds supply," she said. In other words, customers want more AI computing power than Microsoft can currently provide,. Microsoft expects Azure to grow even faster next quarter, by 49% to 50% at constant currency, which means leaving out the effect of exchange rates.

A side note for anyone who saw yesterday's episode: this is the same Microsoft that, earlier in the week, offered voluntary buyouts to around 60,000 employees. Both things are happening at once: heavy spending on AI infrastructure, and a reshaped workforce.

OpenAI reportedly plans to buy Google's AI chips

Pichai's comment about chip demand connects to a report that came out the same day. According to The Information, which cited people familiar with the matter, OpenAI plans to buy AI chips from Google. The report was summarized by Techstrong.ai, and the financial terms were not disclosed. OpenAI and Google both declined to comment.

Some background helps here. Almost all of today's AI runs on specialized chips. The best known are GPUs, made by Nvidia, and OpenAI is one of Nvidia's largest customers. Google makes its own kind, called TPUs, or tensor processing units. It first developed them more than a decade ago to speed up its own search engine, and they have since become the main alternative to Nvidia's chips. Google now sells TPUs directly to some enterprise customers and offers them through partners, including Oracle.

The relationship isn't new. OpenAI already rents Google's chips through Google Cloud, and Techstrong notes that OpenAI recently said it would make its frontier models, its most advanced ones, available through Google Cloud. Buying chips outright would go a step further, and it would deepen ties between two companies that compete directly in chatbots and search.

OpenAI is also building its own chip. Techstrong reports that the project, called Jalapeño, is being developed with the chipmaker Broadcom and is led by Richard Ho, a former Google engineering leader who helped develop early TPUs. OpenAI plans to start deploying its first custom processors late this year.

The common thread is that AI chips are expensive and in short supply, so the companies that need the most of them want more than one source. Alphabet's chief financial officer, Anat Ashkenazi, has said Google expects initial revenue from selling TPUs to outside customers by the end of the year, with most of it coming in 2027.

Google opens Gemini and Search to more apps

Google had a second announcement on Wednesday. As TechCrunch reported, it launched a Connected Apps Platform that lets businesses link their own apps to Gemini and Google Search.

Connected apps first appeared in Gemini and Search earlier this year. The idea is that you link a service you already use, and the AI can take actions in it for you, such as designing a flyer in Canva or building a playlist in YouTube Music, without you opening those apps yourself. This is a small example of what people mean by AI agents: AI that doesn't just answer questions but carries out tasks.

Gemini already has more than 50 connected apps, including Adobe, Experian, Localiza and Peloton, plus Google's own services such as Photos, YouTube Music and Google Home. Search currently offers a smaller set, including Instacart, Canva and Yahoo Fantasy Football, and Google says connected apps are coming to Search more broadly over the next few weeks.

With the new platform, any business or developer can build a connected app, test it and submit it through a tool called the Google Connected Console, and choose whether it appears in Search, in the Gemini app, or both. Developers connect their service using MCP, the Model Context Protocol, a common standard that lets AI assistants talk to other software. After launch, they can track traffic and errors. The meditation app Calm is one of the next partners and will launch on both Gemini and Search soon.

For now, the platform is a public preview for US developers, with a waitlist, and Google did not say when it will be broadly available.

Why does this matter beyond developers? More and more, people may ask an AI assistant to get something done instead of opening an app. TechCrunch points out that ChatGPT and Meta are working on similar ideas, and argues that these platforms only succeed if there's a wide range of apps for people to use.

Anthropic offers US governments 40% off Claude

Meanwhile, Anthropic, the company behind the Claude chatbot, announced a bigger agreement with the US General Services Administration, or GSA. The GSA is the federal agency that handles buying for the government, so a deal with it makes a product easier for agencies to purchase.

Under the new agreement, government customers get 40% off Claude Enterprise, Claude for Government and the Claude API, which is the version developers build their own tools on. The discount isn't limited to federal agencies: state, local, tribal and territorial governments can use it as well. There is no minimum spending commitment, so agencies pay only for what they use. Anthropic estimates the discount could save the government $1 billion.

Anthropic also says that about 500 US federal, state and local government organizations already use Claude, and it shared some examples. In one, Utah gave Claude Code, Anthropic's coding tool, to 34 developers across 10 state agencies for a four-week pilot, and 71% of them reported saving 20 or more hours. In another, Anthropic says a state IT developer moved nearly 900 software functions to a new framework in two days, a project that had been planned for a month. These are Anthropic's own figures and descriptions, not independent results.

The company says it will also launch a program to help agencies put Claude to work, with details in the coming weeks. On security, it says Claude for Government has FedRAMP High authorization, the US government's top tier of cloud security approval for unclassified data.

Trump says he'll ban the term AI

The last story is about a word. On Tuesday, President Trump posted on Truth Social that "in short order" his administration will be officially banning the term "AI/Artificial Intelligence" in all relevant documents, as Variety reported. He called it "a very bad and inaccurate name" and said he prefers "SI," for Super Intelligence. He also asked when The Wall Street Journal would start using the new term.

This builds on something that already exists. An executive order signed on September 29 says the executive branch "shall use the terms 'Super Intelligence' and 'SI'" in place of "Artificial Intelligence" and "AI." What the new post didn't explain is how a wider ban would work, or what it would cover. Gizmodo notes that a ban, as opposed to standardizing the language agencies use, would likely need some kind of enforcement mechanism added to the existing order.

Reactions in the industry have been mixed. According to Variety, Elon Musk's AI business now calls itself SpaceX Super Intelligence, or SpaceXSI, after Musk said he would make the change, although its website is still x.ai. Variety also reports that Microsoft CEO Satya Nadella has used the "super intelligence" wording, though Microsoft hasn't renamed any products. Most AI companies, including OpenAI and Anthropic, still describe themselves with the words "artificial intelligence" and "AI."

The big picture

Put the day's stories side by side and a pattern shows up. The AI boom runs on very physical things: chips, servers and the data centers that hold them. Right now, the companies that own and rent out that hardware are doing very well. Google Cloud doubled in a year, Azure grew 48%, and Microsoft says it still can't supply all the computing power its customers want. Even OpenAI, one of the biggest buyers of AI chips, is reportedly looking for another supplier.

At the same time, the payoff for everyone else is still being worked out. Google is moving ads into its AI answers as growth in classic search slows, Google and others are trying to make AI assistants useful by connecting them to the apps people already use, and AI companies are cutting prices to win large customers such as governments. How that spending turns into everyday products, and who ends up paying for it, is the story to keep watching.