Power and boards carried the sharpest AI numbers of 8 November 2026, by our reading of the news: Anthropic bought commitments for 1.7GW of gas-fired generators for $3.4bn, the Financial Times reported, and Nvidia’s board opened an independent investigation 9 days before its 17 November results.
As we read it, the rest of the day’s coverage priced the build: credit default swaps from 139 to 1,155 basis points, Apple’s AI push, China’s campus economics and two new pushback measures. Each item links to the full piece, which names its sources.
Business Pill · GRID COST ALLOCATION
An evergreen Business Engineer explainer on who pays for new power lines. It teaches the general idea only and does not discuss the companies in this roundup.
AI Daily Briefing · video
Today’s stories in about a minute (8 Nov 2026).
The key insight: As we read it, the day’s coverage put a price on securing AI power: Anthropic bought commitments for 1.7GW of gas-fired generators for $3.4bn, the FT reported, about 99% of a quarter of Caterpillar’s Power Generation sales by our arithmetic, while credit default swaps cost about 8.3 times as much for CoreWeave as for Nvidia. We do not say one explains the other.
Anthropic bought commitments for 1.7GW of gas-fired generators for $3.4bn, the Financial Times reported on 8 November, about 4 times the 420MW of wind farms that Envision’s Galaxy AI campus in Inner Mongolia has built for its own load, according to a SemiAnalysis site report, by our arithmetic. Read: the two ways to buy power.
The $3.4 billion is about 99% of the $3.421 billion of Power Generation sales Caterpillar reported for the third quarter of 2026, by our arithmetic; according to the FT, Anthropic took over generator orders that OnePWR Solutions had placed with Caterpillar two years ago. Caterpillar’s filing puts its firm backlog at approximately $78.1 billion at 30 September. Read: the Caterpillar yardstick.
A City of Starbase notice says SpaceX proposes to construct, own and operate a gas-fired power plant and super-compute facility on an approximately 444-acre tract, with approximately $24 billion of estimated capital investment for initial phases, about $54 million per acre by our arithmetic; the city commission will consider a tax abatement on 9 December. Read: the Starbase notice.
The Senate is scheduled to vote at 5:30 p.m. on Monday, 9 November, on a motion to invoke cloture on the motion to proceed to H.R. 2347, according to the Congressional Record; posts on X and Latitude Media describe it as the test for the bipartisan permitting bill that would make new data centers pay for the grid costs they add. In the UK, growth minister Dan Tomlinson said the UK will not slow down on AI, the FT reported, while data centres queueing for the grid total 73GW, the Guardian reported, about 12 times the at least 6GW of AI-capable capacity the government’s roadmap says is needed by 2030, by our arithmetic. Read: Monday’s vote and the UK queue.
Power capacity bought as equipment: Anthropic bought commitments for 1.7GW of gas-fired generators for $3.4bn, the Financial Times reported, about 4 times the 420MW of wind farms Envision’s Galaxy AI campus has built for its own load, according to SemiAnalysis; the multiple is our arithmetic.
Credit, Priced Name by Name
CoreWeave’s five-year credit default swaps traded around 1,155 basis points on 6 November, according to Deutsche Bank data cited by CNBC, against roughly 139 for Nvidia and roughly 298 for Oracle; insuring CoreWeave’s debt costs about 8.3 times as much as insuring Nvidia’s, by our arithmetic. CNBC reported Nvidia’s swaps up 76% in the last month. Read: the CDS ladder and Nvidia’s 139 basis points.
Goldman Sachs’s credit strategy team tallied AI hyperscaler debt supply at $296 billion so far this year, against the bank’s $250 billion forecast for 2026, MarketWatch reported, about 18% over by our arithmetic. Across six large AI financings we covered, $8.0 billion of $132.3 billion, about 6%, sits in hybrid instruments between debt and equity, by our arithmetic and classification. Read: the $296 billion and the hybrid middle.
Nebius told shareholders that the expected payback on deals signed in its second quarter is “1 year and 10 months”, while it uses “a five-year useful life for our server and network equipment”; 22 months is about 37% of 60, by our arithmetic. A BCG Institute analysis says “roughly $2.6 trillion in annual end-market spending will be needed by 2030” and puts $1.8 trillion of it on enterprise customers. Read: Nebius’s payback and BCG’s $2.6 trillion.
Apollo’s two pending AI chip deals total $75 billion, more than the $44.9 billion of chip-backed financing eight neoclouds have raised since 2023 combined, by our arithmetic from figures compiled by The Information. Read: Apollo’s chip deals.
Apple’s AI Push
Apple CEO John Ternus held a closed-door meeting with software chief Craig Federighi and pushed Federighi to accelerate the company’s AI road map, Bloomberg’s Mark Gurman reported on 8 November. Apple’s fiscal 2026 annual report shows research and development spending of $45.765 billion, up 32%, while net sales rose 15%; R&D reached 9.56% of net sales from 8.30%, by our arithmetic. Read: the Ternus push.
Gurman wrote that Apple “could make Siri more like an agent as early as next year”, and that engineers got six weeks of coding and two of convergence in the first iOS 28 milestone instead of four and four, 50% more coding time by our arithmetic. Mike Rockwell, who led the development of Siri AI, “continues to seriously consider stepping down”, Gurman reported. Read: agentic Siri and Rockwell’s role.
Apple cut its tax bill by $2.7bn this year by channelling earnings into Singapore, more than the $2.6bn it saved via Ireland, the FT reported; by our arithmetic from Apple’s 10-K, the net Singapore saving of $2.67bn equals 24% of the $11.2bn rise in its R&D. Code in the iOS 27.3 beta adds a “Conversational Apps” category for Apple’s $299 HomeView hub, according to the code investigator known as pdfu, as MacRumors reported; Apple has not announced chatbot apps for HomeView. Read: the Singapore saving and the HomeView code.
China’s Build, Measured
SemiAnalysis says Envision’s Galaxy campus in Inner Mongolia delivered its first 120 MW phase within a year at a fifth of US costs, and that the region “has lined up 25GW of announced datacenter projects”. The same report says Chinese data centre rents have “roughly halved over the past five years, to about $40/kW/month excluding power”, and that a live 5MW hall at Galaxy runs 800V DC to the rack. Read: Galaxy’s build and the rents.
China Telecom’s research institute, YOFC and ZTE sent 64 Tbps in each direction over a 1,007.48-kilometre hollow-core fiber link, the Global Times reported, citing CCTV, with one-way transmission over 1,000 kilometres taking 3.43 milliseconds. Read: the fiber trial.
Staff at Tencent, DeepSeek and Alibaba’s Qwen named five distinct caveats on openness in remarks reported between 4 and 8 November, by our count. On the TCSExam benchmark from Tsinghua, Moonshot AI and the Shanghai Qizhi Institute, OpenAI’s GPT-6 Astra produced a machine-checked Lean proof in 70.0% of attempts and resolved none of the 30 open conjectures; Moonshot’s Kimi K3.1 Pro came second on theorem proving at 38.3%. Read: the five caveats and the theorem test.
Boards and Briefings
Nvidia’s board said on 8 November that it is “initiating an independent investigation in light of recent reporting that founder and CEO Jensen Huang has been the target of a foreign intelligence effort”. The review opens 9 days before Nvidia reports third-quarter results on 17 November, and 10 days before the closed-door briefing three House committee chairs have asked Huang to give on 18 November, by our arithmetic. Read: the board’s statement and the House request.
Roelof Botha took a seat on SpaceX’s board in June without first informing Sequoia Capital, according to five people with direct knowledge of the matter cited by the Financial Times; SpaceX’s Form 8-K says its board elected Botha on 16 June 2026. Alphabet’s quarterly report says its holdings include $75.2 billion of SpaceX shares “subject to short-term restrictions on the ability to sell”, 28.4% of its $265.1 billion of cash and marketable securities, by our arithmetic. Read: the Sequoia seat and Alphabet’s SpaceX stake.
Stocks and Pushback
Micron’s shares closed at $690.00 on 6 November, up 141.76% from the end of 2025, while the simple average change for Amazon, Microsoft, Meta and Alphabet was +1.4%, by our arithmetic from Nasdaq.com closes. Intel’s shares closed at $56.28, down 53.2% from 30 September, a fall in market value of roughly $340 billion, by our arithmetic. Read: the year’s split and Intel’s quarter.
The Bloomington, Indiana, Common Council voted on 4 November to ban the distribution and deployment of superintelligent AI inside the city, the Indiana Daily Student reported; we checked a viral post’s five claims against the ordinance, and two match as written, by our count. New York Assemblymember Alex Bores wrote on X: “Congress should ban RSI, this year if possible”. Read: the ordinance, the claim check and Bores’s post.
Apple’s research and development spending and net sales, fiscal 2025 = 100: R&D 132 and net sales 115 in fiscal 2026, our index from the 32% and 15% rises in Apple’s annual report; R&D reached 9.56% of net sales, from 8.30%.
The Structural Read
As we read it, power arrived as equipment orders: gas-fired generators in the Anthropic case and a gas-fired plant in SpaceX’s Starbase notice, against wind farms sized above the load at China’s Galaxy campus.
As we read it, credit priced the builders one by one: Goldman’s tally of $296 billion of hyperscaler debt this year, against a $250 billion forecast, sat beside swaps ranging from roughly 139 to about 1,155 basis points.
As we read it, the rest of the day’s reporting was about who governs the build: Nvidia’s board review, Apple’s leadership push on AI, and new pushback measures in Bloomington and in Congress. We do not say why prices moved.
Dan Tomlinson, UK growth minister, on the pace of AI investment, according to a Financial Times report as quoted in our piece of 8 November 2026
“too low for what we need to stay at the forefront as AI technology develops”
Three Implications
AI BUILDERS Anthropic bought commitments for 1.7GW of gas-fired generators for $3.4bn, the FT reported, about 99% of Caterpillar’s third-quarter Power Generation sales by our arithmetic.
LENDERS CoreWeave’s five-year swaps traded around 1,155 basis points against roughly 139 for Nvidia, about 8.3 times as much, by our arithmetic.
APPLE’S SHAREHOLDERSApple’s research and development spending rose 32% to $45.765 billion in fiscal 2026 while net sales rose 15%, according to its annual report.
The analysis describes a stack of nine layers, “running from energy at the bedrock to governance at the ceiling”, and says: “The supercycle has a ceiling now, and the ceiling is political.”
As we read it, the day’s generator, plant and wind items sat on the Energy & Physical layer of that stack, the credit prices on the Compute Capacity layer, and Nvidia’s board review and the Bloomington ordinance on the governance layer. We do not say one caused the other. The analysis was published in June 2026 and does not discuss these events.
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 threads, and the reading of the day as a power and governance story, are our grouping, not a finding.
We do not claim that any one story moved prices or caused another. The generator figure compares a purchase with a quarter’s sales; the CDS levels come from different data providers; several items rest on outlet reports citing people familiar with the matter or on posts on X; and Nvidia’s board statement does not say what the investigation will find.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
For anyone pricing the AI build this week, two numbers sit side by side: 1.7GW of gas-fired generators bought for $3.4bn, about 99% of a quarter of Caterpillar’s Power Generation sales by our arithmetic, and credit default swaps that cost about 8.3 times as much for CoreWeave as for Nvidia.
For the frame behind the day, read The AI Supercycle on The Business Engineer, on the nine layers from energy to governance: read the analysis.
A note on sourcing. This roundup draws on FourWeekMBA’s news coverage of 7 and 8 November 2026, linked above; every story was built from its own primary source, which it cites and links. We re-read each piece on 8 November 2026 and took every figure here from it. The 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.
