Google’s and Microsoft’s quarters were the day’s biggest thread on 28 October 2026, by our grouping of the news. Alphabet said Google Cloud revenue rose 103% to $30.7 billion, while it spent $51.4 billion on property and equipment against $42.3 billion of cash from operations. Microsoft said Azure revenue grew 48%, and its quarterly report puts additional leases that had not yet commenced, primarily for datacenters, at $364.6 billion.
The threads below are Google’s quarter, from the filings; Microsoft’s quarter; who pays for the build; suppliers and deals; labs and models; and rules, places and work. Each item links to the full piece, where the primary source is cited. This roundup selects from the day’s coverage.
Business Pill · FREE CASH FLOW
An evergreen Business Engineer explainer on free cash flow, the cash a company has left after capital expenditures. It explains the general idea and does not discuss the companies in this roundup.
The key insight: As we read it, the day’s filings put AI spending ahead of cash at Google: Alphabet spent $51.4 billion on property and equipment against $42.3 billion of cash from operations, while Google Cloud revenue rose 103% to $30.7 billion.
Alphabet said Google Cloud revenue rose 103% to $30.7 billion in the third quarter, from $15.2 billion a year earlier, with Cloud operating income of $11.1 billion against $3.6 billion. Its capital expenditures reached $51.4 billion, against $24.0 billion a year earlier, and free cash flow was negative $9.1 billion by Alphabet’s own reconciliation. Read: Google’s quarter. By our arithmetic from the two cash flow statements, Alphabet’s capital spending equalled 121% of its operating cash flow in the quarter, and Microsoft’s equalled 68%. Read: the cash comparison.
Alphabet held $13.3 billion of inventory on 30 September, up from $2.4 billion at the end of 2025, and its second-quarter 10-Q says that inventory “consists primarily of hardware related to TPU systems for sale to enterprise customers and devices”. Read: the TPU inventory. Google Cloud chief executive Thomas Kurian named Apple and Anthropic among the companies using Google’s TPU chips on the earnings call, according to a caption transcript of the call.
Read: the TPU customers. Alphabet’s backstops for other companies’ data-center payment obligations had a notional value of $43.8 billion on 30 June 2026, up from $0 on 31 December 2024. Read: the backstops.
Alphabet spent nothing on repurchases of its stock in the third quarter, against $11.5 billion in the same quarter of 2025. Read: the buybacks. Its “Alphabet-level activities” posted an operating loss of $7.7 billion, against $4.5 billion a year earlier, and the release says these “primarily reflect expenses related to our shared AI research and development”. Read: shared AI R&D. Other Bets had an operating loss of $2.0 billion on $372 million of revenue, a loss of $5.50 for every $1 of revenue by our arithmetic.
Read: Other Bets. Alphabet recorded a net loss of $8.7 billion on equity securities, which its release says reduced diluted earnings per share by $0.55. Read: the stake loss.
Revenue from customers in the United States rose 39% to $67.7 billion; by our arithmetic, the US supplied 76% of the $24.8 billion Alphabet added in a year, excluding hedging. Read: the US share. Google Search & other revenue rose 12% to $63.4 billion, the slowest year-over-year growth for the line since April to June 2025 by our arithmetic, while Alphabet said its Search query growth rate was the highest in the last five years.
Read: Search revenue. Alphabet said its first-party model APIs are processing nearly 32 billion tokens per minute, more than four times a year ago. Read: the token count. It had 204,757 employees at the end of September, against 190,167 a year earlier. Read: the headcount.
Alphabet’s third-quarter 2026 cash from operations ($42.3 billion) and purchases of property and equipment ($51.4 billion), from its 28 October release as our pieces report them. The negative $9.1 billion is Alphabet’s own free cash flow figure.
Later on 28 October, Alphabet’s chief financial officer Anat Ashkenazi said on the earnings call that it had future purchase commitments of $924 billion at the end of September, against a Google Cloud revenue backlog of $561 billion; by our arithmetic, that is $1.65 of commitments for every $1 of backlog. Read: the commitments. Thomas Kurian said on the call that Google Cloud now has 18 product lines “with more than $1 billion each in annual revenue.” Read: the product lines.
Microsoft’s Quarter
Microsoft reported that Azure revenue grew 48% to $33.2 billion in the quarter to 30 September, its first quarter of fiscal 2027. Cash spent on property and equipment was $39.4 billion, against $19.4 billion a year earlier, and by our arithmetic free cash flow fell to $18.6 billion from $25.7 billion. Read: Microsoft’s quarter.
Microsoft’s quarterly report says it recorded revenue from commercial arrangements with OpenAI, inclusive of revenue-sharing payments, of $10.0 billion for the three months, about 10.8% of its revenue by our arithmetic. Read: OpenAI’s share. The commercial portion of its remaining performance obligation was $677 billion on 30 September, against $678 billion on 30 June. Read: the backlog.
The same report lists additional leases, primarily for datacenters, that had not yet commenced of $364.6 billion, about 5.7 times Microsoft’s $64.5 billion of cash, cash equivalents and short-term investments by our arithmetic. Read: the leases. Depreciation expense was $11.1 billion for the three months, against $7.1 billion a year earlier. Read: the depreciation bill. Two explainers walk through how Google makes money now and how Microsoft makes money now.
Who Pays for the Build
Bloomberg reported that financial firms, companies and governments have raised a combined $6.61 trillion through investment-grade and junk-bond sales so far this year, up 15%, according to data compiled by Bloomberg. By our sum, Alphabet took in $147.2 billion from issuing debt, common stock and preferred stock in the nine months to 30 September 2026. Read: the bond record. Bloomberg also reported early signs that the credit market is “approaching the limits of how much more debt it can absorb”. Read: AI credit.
CoreWeave’s balance sheet at 30 June 2026 lists $5.52 billion of cash and cash equivalents against $35.07 billion of debt, about 16 cents of cash for every $1 of debt by our arithmetic. Read: the neocloud balance sheets. Oracle said it expects free cash flow to turn positive in the second half of fiscal 2028 and set a target of at least $30 billion for fiscal 2029; free cash flow was negative $23.7 billion in fiscal 2026. Read: Oracle’s plan.
Anthropic’s S-1 says that after 30 June it entered into agreements with aggregate future minimum payment obligations of approximately $245 billion, about 1.9 times the $131.7 billion of cash and short-term investments it held on 30 June by our arithmetic. Read: Anthropic’s compute. The S-1 gives Anthropic the ability to issue up to $42.0 billion of convertible notes to Broadcom or a designated financial partner, solely to finance quarterly rent, a structure Michael Burry wrote about in a public Substack note. Read: the Broadcom notes.
The Federal Reserve held its benchmark rate at 3.75% to 4% in a 10-2 vote in which two officials wanted to raise it; its statement says “Productivity growth is strong, and capital investment is robust”. Read: the Fed. ING estimates that artificial intelligence accounts for about a third of US economic growth in 2026. Read: ING’s estimate.
Suppliers and Deals
SK hynix said it earned 83.3338 trillion won in third-quarter operating profit, an operating margin of 80%, and that long-term agreements with customers now cover more than 50% of its production capacity for 2027 and 2028. Read: SK hynix. Amphenol booked record orders of $12.2 billion as sales rose 65% to $10.2 billion. Read: Amphenol. Vertiv said organic orders grew more than 100% and raised its full-year 2026 net sales guidance by $650 million. Read: Vertiv.
KLA said it expects December-quarter revenues of $4.7 billion, plus or minus $200 million; by our arithmetic, the midpoint is 42.5% above its December 2025 quarter. Read: KLA. Equinix reported record annualized gross bookings of $533 million, up 35%, and by our arithmetic spent 60.6% of its revenues on purchases of other property, plant and equipment, against 42.7% at Microsoft and 40.3% at Alphabet on the same basis.
Read: Equinix. ServiceNow raised its 2026 target for AI annual contract value to $1.7 billion, from $1 billion at the start of the year. Read: ServiceNow.
Bloomberg reported that Marvell Technology is in advanced talks to buy Lumentum Holdings, according to people familiar with the matter; by our arithmetic, Bloomberg’s market value for Lumentum is about 27.5 times its revenue for its fiscal year to 27 June 2026. Read: Marvell and Lumentum. Reuters reported that the US Department of Justice has made inquiries about how Nvidia’s $12.93 billion deal to buy Hugging Face would affect competition, citing two people familiar with the matter. Read: the DOJ questions.
Labs and Models
Google said Gemini 4 Carbon “ranks #1 on the recently updated v5.0 of Artificial Analysis Intelligence Index at 59”, and that it will launch at $2 per million input tokens and $10 per million output tokens; by our arithmetic, that is half Anthropic’s listed price for Opus 5.5. Read: Gemini 4 Carbon. Artificial Analysis said Claude Opus 5.5 (max) leads at 58 on version 5 of its index, ahead of GPT-6.1 Sol (max) at 55, which by our arithmetic is about 7.9 times cheaper per task. Read: the AI Index.
Anthropic said in a research post that it used “an unreleased, internal research model” to find critical vulnerabilities, and that 62 Chrome CVEs have been assigned and patched as of Chrome 156. Read: the exploit chains. Anthropic said it has partnered with 20 scientific foundations, institutions and research groups in an expansion of its AI for Science program. Read: AI for Science. Similarweb said Claude recorded 1.101 billion visits over the past 28 days. Read: Claude’s traffic.
Rebuilding a Bloomberg comparison from SEC filings, our arithmetic puts Anthropic’s 2025 revenue at about 112 times its 2023 revenue, against a comparable multiple of about 40 times for Nvidia. Read: the pre-IPO growth. Business Insider reported that Cognition has received investor offers valuing it at between $70 billion and $80 billion, citing people familiar with the matter. Read: Cognition.
Rules, Places and Work
The Effingham Herald reported that plaintiffs challenging OpenAI’s proposed Project Camellia data center in Effingham County, Georgia, have asked a judge to rule in their favor without a trial; according to the plaintiffs’ filing, as the Herald reports it, one agreement included nearly $900 million in tax abatements. Read: the Georgia suit. The Guardian reported that real estate agents have been offering to buy houses near xAI’s gas-turbine power plant in Southaven, Mississippi, and that a realtor told one resident there is a target to purchase 418 homes.
Read: xAI’s plant. The UK Treasury published a Data Centre Covenant as part of Budget 2026, alongside an AI package worth £700 million. Read: the UK plan.
New York Gov. Kathy Hochul signed a bill that bars AI chatbot operators from offering a list of “unsafe AI companion features” to users in the state unless they have first determined, by permitted age-assurance methods, that the user is not a minor; the attorney general can seek civil penalties of up to $25,000 per violation. Read: New York’s law. Three House Democratic committee leaders asked Anthropic, Google, Meta, OpenAI and SpaceXSI to report certain AI incidents “within 48 hours of discovery”. Read: the incident letters.
The parents of a 31-year-old Amazon software engineer have sued OpenAI, alleging that ChatGPT practiced medicine without a license, The New York Times reported; these are allegations, and no court has ruled on them. Read: the lawsuit. McKinsey, BCG and Bain have raised US starting base salaries for hires joining in 2027, according to Management Consulted, which counts it as the first starting salary increase at the three firms since August 2022. Read: consulting pay.
Capital spending as a share of quarterly revenue on the same basis, as our Equinix piece reports it. The ratios are our arithmetic from each company’s filings.
The Structural Read
Alphabet held $13.3 billion of inventory on 30 September, up from $2.4 billion at the end of 2025, and its backstops for other companies’ data-center payment obligations had a notional value of $43.8 billion on 30 June 2026.
Microsoft’s quarterly report lists additional leases, primarily for datacenters, that had not yet commenced of $364.6 billion, and puts revenue from commercial arrangements with OpenAI at $10.0 billion for the three months.
As we read it, the financing coverage sat beside the earnings: Bloomberg reported that bond sales so far this year have reached a combined $6.61 trillion, and CoreWeave’s balance sheet lists about 16 cents of cash for every $1 of debt by our arithmetic.
Giordano Albertazzi, Vertiv chief executive, 28 October 2026, as quoted in our piece
“We are in the middle of the largest infrastructure buildout and it’s accelerating”
Three Implications
HYPERSCALERS By our arithmetic from the two cash flow statements, Alphabet’s capital spending equalled 121% of its operating cash flow in the quarter, and Microsoft’s equalled 68%.
DATA CENTER LANDLORDS By our arithmetic, Equinix spent 60.6% of its revenues on purchases of other property, plant and equipment, against 42.7% at Microsoft and 40.3% at Alphabet.
LENDERS AND BOND BUYERS Bloomberg reported early signs that the credit market is “approaching the limits of how much more debt it can absorb”.
The analysis says: “The $1T capex cycle is being underwritten by a $1T concentrated demand commitment from two counterparties whose combined free cash flow is deeply negative.”
As we read it, the day’s filings show the spending side of that cycle, in Alphabet’s capex and Microsoft’s unstarted leases. We do not say one caused the other. The analysis was published in May 2026 and does not discuss these quarters.
What Is Not Established
This roundup adds no new reporting and selects from the day’s coverage. Every figure above is taken from the linked FourWeekMBA piece, each built from its own primary source. The six threads are our grouping, not a finding.
We do not claim that these stories are connected or that one caused another. The capital spending ratios compare companies on the same basis but are our arithmetic, not company metrics. Kurian’s naming of TPU customers rests on a caption transcript of the call. The Marvell talks, the DOJ inquiries and the Cognition offers rest on reports citing people familiar with the matter. The OpenAI lawsuit makes allegations that no court has ruled on.
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 said on 28 October that Google Cloud revenue rose 103% to $30.7 billion, while its $51.4 billion of capital expenditures exceeded its $42.3 billion of operating cash flow. Microsoft’s quarterly report puts its additional leases not yet commenced, primarily for datacenters, at $364.6 billion.
The Business Engineer’s AI Capex Map reads the financing side of the build: read the analysis.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. This roundup draws on FourWeekMBA’s news coverage of 27 and 28 October 2026, linked above; every story was built from its own primary source, which it cites and links. We re-read each piece on 28 October 2026 and took every figure here from it. The six threads and the figures in our charts are our own grouping and selection. Nothing here is a forecast, and nothing here is financial or investment advice.
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.
