AI Industry Daily Briefing — November 2, 2026
Alphabet's filing shows Google Cloud revenue up 103% to $30.7 billion and free cash flow of negative $9.1 billion, Microsoft reports commercial backlog of $677 billion, and Google says its test satellite returned the first AI response from a data-center-class chip in orbit.
The Executive Read
The last week of earnings gave the AI buildout its first full set of receipts, and they point two ways at once. Demand is visible in the filings. Alphabet reports Google Cloud revenue of $30.7 billion, up 103% from a year earlier, and says its own model interfaces now process nearly 32 billion tokens a minute, a token being a small chunk of text a model reads or writes. Microsoft reports that contracted future cloud revenue, which it calls commercial remaining performance obligation, rose 73% to $677 billion. The cost is visible too. Alphabet spent $51.4 billion on equipment in one quarter and ended it with negative free cash flow of $9.1 billion, and Microsoft’s finance chief told investors that outside suppliers’ delays will push some capacity into the first half of 2027. Everything in this edition rests on company filings or company posts, which are the companies’ own accounts. We found no independent measurement of the usage figures. Two smaller items round out the day: Google says its test satellite has returned the first answer from a data-center-class chip in orbit, and a South Korean chip startup says it raised $620 million.
Top AI Headlines
Alphabet’s Filing: Cloud Revenue Doubles, Cash Goes Negative
What happened. Alphabet’s earnings release, filed with the Securities and Exchange Commission on October 28 for the quarter ended September 30, reports revenue of $127.6 billion, up 25% from $102.3 billion. Google Cloud revenue was $30.7 billion, up 103% from $15.2 billion, and Cloud operating income was $11.1 billion against $3.6 billion a year earlier. The release says sales of Google’s TPU systems, its in-house AI chips, contributed meaningfully. Total operating income was $40.6 billion, a 31.8% margin. Net income available to common stockholders fell to $26.2 billion from $35.0 billion, which the release attributes mainly to an $8.7 billion loss on equity securities, an accounting item outside operations. Purchases of property and equipment were $51.4 billion, up from $24.0 billion, and free cash flow was negative $9.1 billion. Alphabet’s first-party model interfaces process nearly 32 billion tokens a minute, more than four times the rate a year earlier, according to the release. It says the Gemini app has reached one billion users. CEO Sundar Pichai said the company is “now on a half-trillion dollar revenue run rate” and that “Gemini 4 Argon and Carbon are now leading at the frontier.”
Why it matters. Two numbers move together here: Cloud revenue doubled, and cash flow turned negative. Alphabet is spending faster than it earns, even at 31.8% margins. The usage figures, tokens and users, are Alphabet’s own, and the release does not define how a “user” or a token is counted. The “leading at the frontier” line is the CEO’s claim, not a benchmark result. CNBC reported that Alphabet shares fell about 10% in extended trading after the report and that the company held its 2026 capital spending guidance at $195 billion to $205 billion; the 10% figure is CNBC’s, not Alphabet’s.
Business implication. For companies buying AI capacity, the message is that demand for TPU-based systems is strong enough that Google sells them as products. For suppliers and lenders, it is that even the largest cloud companies are now funding growth with outside capital or by drawing down cash. This is informational, not investment advice.
Sources: Alphabet, Q3 2026 results exhibit 99.1 (SEC, Form 8-K); CNBC, “AI buildout hits snag as Google and Microsoft put off raising capex forecasts” (October 29).
Microsoft: $677 Billion in Contracted Cloud Revenue, and a Capacity Delay
What happened. Microsoft’s results for the quarter ended September 30 show revenue of $92.3 billion, up 19%, and operating income of $45.9 billion, up 21%. Azure revenue grew 48% to $33.2 billion. Microsoft Cloud revenue was $63.9 billion, up 30%. Commercial remaining performance obligation, the value of signed contracts not yet recognised as revenue, rose 73% to $677 billion. GAAP net income was $36.6 billion; non-GAAP net income, which strips out gains tied to Microsoft’s OpenAI investments, was $36.4 billion. Those gains added $207 million to net income this quarter. CNBC reports that Microsoft cut its fiscal-year capital spending forecast to $170 billion from $175 billion, and that finance chief Amy Hood said delays at third-party providers of graphics processors will push some capacity into the first half of 2027.
Why it matters. A backlog is a promise to pay, not cash received. Microsoft’s release does not say how much of the $677 billion comes from OpenAI or any other single customer, and we did not find a breakdown. The capacity delay matters because, on CNBC’s account of Hood’s remarks, supply rather than demand is limiting how fast Microsoft can serve customers. Microsoft is also betting on its own chips. On October 12 its blog announced the Maia 300, an inference chip, with production deployment in 2027; we cover it under Model and Product Updates.
Business implication. Buyers planning large Azure commitments should assume tight capacity into mid-2027 on Microsoft’s own account. CNBC attributes the lower spending forecast to Hood’s comments about supplier delays; the release itself does not explain the cut. This is informational, not investment advice.
Sources: Microsoft Investor Relations, FY27 Q1 earnings release; CNBC (October 29), which quotes Hood and reports the capital-spending cut.
Google Says Its Orbiting TPU Has Answered Its First Prompt
What happened. In an October 30 post, Google says its Project Suncatcher prototype satellite reached what it calls a “Hello World” milestone: the first AI response from a data-center-class TPU in orbit. The satellite launched on October 1. During a ground-station pass, the team sent a prompt to the onboard Gemma 4 model, an open model Google released, asking for a haiku about being in space. The reply came down as the satellite passed Svalbard, Norway. Follow-up tests included limericks and telemetry prompts reporting the status of the payload.
Why it matters. The idea is to run AI computing in orbit, where sunlight is steady and cooling works differently. This test shows a chip can boot, accept a prompt and return text. The post gives no performance figures, response times or power data, and the thermal and radiation tests that would show whether the chips survive are still ahead. All of it is Google’s own report from a single prototype.
Business implication. Nothing here changes data-center economics today. Google’s stated next step is a pair of interconnected satellites in 2027. Treat the project as research until it publishes measurements.
Model and Product Updates
Microsoft’s Maia 300 targets inference. Microsoft’s October 12 blog post, by cloud chief Scott Guthrie, describes the Maia 300, its third-generation AI accelerator. Compute chiplets use TSMC’s N2P process. Each chip has eight HBM4 memory stacks (a fast memory type) with 18.0 terabytes a second of peak bandwidth in a 950-watt power envelope. A rack holds 96 accelerators in a 167-kilowatt envelope with 27.6 terabytes of memory. These are design specifications. Microsoft says it “expects” 40% more tokens per megawatt than leading GPU-based systems and up to seven times the rack-level throughput of Maia 200, its prior chip. Those are projections against unnamed competitors, and the chip is not yet in production. Microsoft says Maia 200 already runs production workloads including Copilot, and that Maia 300 production deployment starts in 2027. A megawatt is roughly the power for several hundred homes.
Google’s Gemini 4 models, by Google’s account. Pichai’s quote in the Alphabet release calls Gemini 4 Argon and Carbon “leading at the frontier.” We did not find an independent benchmark in the filing and treat the phrase as the company’s description.
Sources: Microsoft, “Microsoft to expand Azure Maia system to multi-gigawatt deployments” (October 12); Alphabet, Q3 2026 exhibit 99.1 (SEC).
Regulation and Policy Watch
EU AI Act: the Commission’s own timeline. The European Commission’s AI Act page, last updated October 19, lists the current dates. The “AI Omnibus,” a package that postponed parts of the law, entered into force on July 27 after political agreement on May 7. Under it, the high-risk rules for stand-alone uses such as hiring, education and biometrics apply from December 2, 2027. Rules for high-risk AI built into regulated products such as lifts and toys apply from August 2, 2028. The transparency rules, which require telling people when they talk to a chatbot and labelling AI-generated content, began applying in August 2026. A new ban on AI that generates non-consensual sexual imagery or child sexual abuse material applies from December 2026, the page says. It also lists the first harmonised standard supporting the Act, published in the Official Journal on October 19. A harmonised standard is a technical specification that companies can follow to show compliance.
What we did not find. We searched for new US federal action on children’s chatbot safety and on state-law preemption dated since October 30 and found no new action, only earlier bills and New York’s law covered in recent editions. China’s draft export-control catalogue, covered October 25, remains open for comment until November 15; we found no change to it.
Sources: European Commission, AI Act regulatory framework page (updated October 19).
Emerging Startup Radar
FuriosaAI raises a reported $620 million. CRN Asia reports that the South Korean AI-chip designer closed a Series D of $620 million at a post-money valuation of about $2.9 billion, and that the money will go to scaling production of its RNGD inference chip and developing a third-generation chip. CRN Asia also reports the state-backed National Growth Fund is the largest investor. We could not open FuriosaAI’s own announcement or the Datacenter Dynamics report, and Korean coverage gives the amount as 820 billion won, which is consistent. Bloomberg reported in January that the company planned to raise $300 million to $500 million ahead of a listing; we have not seen Bloomberg’s article ourselves, only a search listing of it. Treat the round as reported, not confirmed by the company in anything we read.
Sources: CRN Asia, “FuriosaAI to scale RNGD inference chip production following Series D funding”. Single trade-press source; no company release read.
AI Infrastructure and Market Signals
Capital spending: the arithmetic. From the filings and CNBC: Alphabet spent $51.4 billion on equipment in the quarter, against $24.0 billion a year earlier. CNBC reports Meta raised the low end of its 2026 range to $135 billion to $145 billion. CNBC also says shares of CoreWeave, Nebius, IREN and Oracle fell after Microsoft’s and Alphabet’s reports, which CNBC ties to higher borrowing costs and debt concerns. CNBC’s line that hyperscaler spending “is expected to top $1 trillion next year” carries no named source in the article, and the Bloomberg link behind it points to a different projection of $1.2 trillion from Goldman Sachs. We treat both as estimates, not facts.
Backlog versus build time. Microsoft’s $677 billion of contracted revenue and Alphabet’s cloud growth both describe demand that has to be served with chips and power that are not yet installed. Hood’s comment about delayed capacity is the clearest on-the-record sign in this week’s coverage that supply is limiting growth at the margin.
Market backdrop. Yesterday’s edition reported that the Nasdaq Composite fell 2.9% on Friday, October 30; we have not found a primary-source reading for today’s session and report none.
Sources: CNBC (October 29); Alphabet 8-K exhibit (SEC); Microsoft IR.
Public Investment Watchlist
Informational only; this is not investment advice.
- Alphabet (GOOGL). Reported Cloud revenue of $30.7 billion, negative free cash flow of $9.1 billion and 2026 capital spending guidance of $195 billion to $205 billion. CNBC reports shares fell about 10% in extended trading on the report.
- Microsoft (MSFT). Commercial backlog of $677 billion; capital spending forecast trimmed to $170 billion, per CNBC. Microsoft’s preliminary bond prospectus was covered in yesterday’s edition.
- Anthropic (proposed Nasdaq listing, ticker ANT per reports). The company published a post on October 26 saying it filed a registration statement; the S-1 is on the SEC’s site. The document was too large for our tools to open in full today, so we report no figures from it, and numbers circulating elsewhere are unverified here. Anthropic’s S-1 on sec.gov.
- Neoclouds (CoreWeave, Nebius, IREN). CNBC reports all three fell after the Microsoft and Alphabet reports.
Watchlist
- November 15: China’s comment period on its draft export-control catalogue closes (MOFCOM Announcement No. 48).
- December 2: The EU’s ban on AI that generates non-consensual sexual imagery or child abuse material begins to apply, per the Commission’s page.
- Anthropic S-1 amendments: Offering size and price had not been disclosed in the coverage we read; later amendments would carry them.
- Maia 300: Microsoft says more on its chip strategy comes at Ignite 2026; production deployment is slated for 2027.
- Suncatcher: Google’s thermal and radiation results from orbit, which it says are next.
A note on today’s reporting: OpenAI’s newsroom and help pages returned errors to our tools, so we could not verify reports of a shared ChatGPT and Codex usage limit or of the GPT-6.1 models, and we have left them out. Several lines above rely on CNBC’s relay of company statements; where that is so, we say it.