Chamath Palihapitiya wrote on X on 1 November that “Power is THE binding constraint in AI/SI. Few really have it.” He was quoting Elon Musk’s post that “There will be a total of around 1.2M Blackwells at Colossus 2 by end of year”.
The risk-factor sections we counted put power inside a bigger bundle. Anthropic’s S-1 gives its single longest risk factor, 4,066 words or 6.1%, to chips, electrical power and data-centre capacity together. Across nine filings, risk factors led by power alone total 623 words, about 0.3%, by our count. Anthropic’s one bundled factor is about 6.5 times that, by our arithmetic.
Business Pill · Financial Conditions
A short explainer on financial conditions: how credit spreads and borrowing costs feed through the economy. It teaches the general idea only and says nothing about the companies in this piece.
The key insight: As we read Chamath Palihapitiya’s post next to the filings, the disagreement is about emphasis: he calls power THE binding constraint, while Anthropic’s longest risk factor, 4,066 words by our count, bundles chips, electrical power and data-centre capacity together, and power-led risk factors across nine filings total 623 words.
His post builds the case as a chain. “Data centers don’t work without power.” “Chips can’t operate profitably unless installed in a functional data center. If you can install chips, it’s because you’ve solved the power problem.”
He stresses timing: “This is not about promises of power tomorrow. Or forecasted power in 2028. Power TODAY.” And he ends: “Solving power allows you to disproportionately shape AI/SI right now.”
Words of risk-factor text by our count: Anthropic’s compute-related risk factors (8,868 words, 13.4% of its S-1 risk section), its longest single risk factor on chips, power and data-centre capacity (4,066 words, 6.1%), and risk factors led by power alone across nine filings (623 words, about 0.3% of 196,067).
What Was Posted on X
He who shapes power, data centers, chips and models into an economically feasible endpoint, disproportionately shapes AI/SI right now.
Why
Power is THE binding constraint in AI/SI. Few really have it.
Data centers donโt work without power. This is not about promises of power tomorrow. Or forecasted power in 2028. Power TODAY.
Chips canโt operate profitably unless installed in a functional data center. If you can install chips, itโs because youโve solved the power problem.
Models arenโt economically viable unless run on cost effective infrastructure which requires a functional data center with low input costs and reliable management. This canโt happen without power.
AI/SI ROI doesnโt pencil unless models generate profitable outputs which necessarily needs functional data centers as the critical input. Again, power.
Ergo, he who can shape data centers, chips and models together into a tightly coordinated stack is someone who has solved power.
Solving power allows you to disproportionately shape AI/SI right now.
Chamath Palihapitiya’s post on 1 November 2026, quoting Elon Musk. He writes that “Power is THE binding constraint in AI/SI” and that “Solving power allows you to disproportionately shape AI/SI right now.” This piece sets his view against FourWeekMBA’s word counts of risk-factor sections.
@XFreeze There will be a total of around 1.2M Blackwells at Colossus 2 by end of year
Elon Musk’s post on 31 October 2026, which Chamath quoted: he wrote that there will be around 1.2M Blackwells at Colossus 2 by the end of the year.
What the Filings Say
Anthropic’s S-1 lists 88 risk factors in about 66,225 words of body text, by our count, as we reported. Its compute-related risk factors run to 8,868 words, or 13.4% of the section, our arithmetic.
We also split the risk sections of nine filings, the latest annual reports of Alphabet, Amazon, Microsoft, Meta, Apple, Oracle, Tesla and CoreWeave plus Anthropic’s S-1, into 382 risk factors and 196,067 words, as we reported. Each risk factor gets one theme in our method, so one that covers chips, power and data centres together counts once, under compute.
That is why power as a stand-alone theme shows 623 words in total. As we read it, Anthropic’s longest risk factor shows how power can be counted inside a compute bundle rather than as a power-led risk factor.
The Power Behind Colossus 2
Musk’s 1.2 million Blackwell figure covers the whole Colossus 2 campus by year-end, as we reported. The gas-turbine plant behind it is the subject of an NAACP lawsuit: a court order lists at least 27 turbines on the site as of 7 April, with a combined capacity of 495 MW, by our arithmetic from its figures, as we reported.
The Business Lever: Chamath’s Chain and the Filings
Alphabet, Amazon, Meta and Microsoft have guided 2026 capital spending that adds up to about $730 billion at the midpoints, we calculated. The IEA says orders for new gas-fired power plants reached 130 GW in 2025, a 25-year high, “with US data centre demand a major driver”.
As we read Chamath’s post next to the filings, his chain of power, data centres, chips and models matches how Anthropic writes its longest risk factor: one bundle of inputs behind compute. The difference is emphasis. He names power first; the filings name it inside the compute supply chain.
The Structural Read
As we read the S-1 counts, Anthropic’s compute-related risk factors run to 8,868 words, about 13.4% of its risk section, more than twice the length of its longest single risk factor.
As we read the method behind the nine-filing count, a risk factor that covers chips, power and data centres together counts once, under compute, which keeps the stand-alone power theme at about 0.3% of the words.
As we read Musk’s post alongside the court order on the plant behind Colossus 2, the 1.2 million Blackwell figure sits next to at least 27 gas turbines with a combined 495 MW, by our arithmetic from the order’s figures.
Chamath Palihapitiya on X, 1 November 2026
“Ergo, he who can shape data centers, chips and models together into a tightly coordinated stack is someone who has solved power.”
Three Implications
FOR READERS OF RISK SECTIONS By our count, Anthropic’s single bundled risk factor on chips, power and data-centre capacity is about 6.5 times all power-led risk factors across nine filings.
FOR THE COLOSSUS 2 BUILD Musk’s post gives about 1.2 million Blackwells by year-end; the court order lists at least 27 turbines with a combined 495 MW, by our arithmetic.
FOR THE CAPEX PICTURE Alphabet, Amazon, Meta and Microsoft’s 2026 guidance adds up to about $730 billion at the midpoints, our arithmetic, while the IEA counts 130 GW of gas-plant orders in 2025.
The framework says “It is a sovereign-scale financing problem dressed in tech-company clothing โ and the financial architecture is starting to invert.”
As we read it, power is one of the inputs that make the build-out a financing problem at that scale. The framework predates Chamath’s post and does not discuss it.
Chamath’s post is his view and cites no data. Our word counts measure how much text a filing gives a risk, not how likely or costly that risk is, and our one-theme method assigns bundled risk factors to compute.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Chamath Palihapitiya calls power “THE binding constraint in AI/SI”. In the risk sections we counted, power sits inside a bundle: Anthropic’s longest risk factor, 4,066 words, covers chips, power and data-centre capacity together, while power-led risk factors across nine filings total 623 words, by our count.
A note on sourcing. Chamath Palihapitiya’s and Elon Musk’s words come from their own posts on X (1 November and 31 October 2026). All word counts are FourWeekMBA’s own counts of the risk-factor sections, published earlier; they measure text length, not the likelihood or cost of a risk.
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.
