Anthropic Did Not Just Buy Power. It Bought Its Place in Line.

2026-11-08T13:35:52Z

1. Anthropic Did Not Just Buy Power. It Bought Its Place in Line. (Investments)

Emerging data

Imagine ordering a powerful computer, only to discover that the electricity needed to run it will take another two years to arrive.

That is becoming a serious problem for the artificial intelligence industry. Companies can raise billions, purchase chips and build data centres, yet still struggle to obtain enough electricity.

On 8 November 2026, the Financial Times reported something important.

Our database points to a bigger connection.

Five developments connect.

Report A: AI companies need electricity on a scale that ordinary commercial buildings rarely require.

Report B: Generator manufacturers face strong demand and long manufacturing schedules.

Report C: OnePWR had reportedly already secured equipment commitments before Anthropic took them over.

Report E: Investors have recently sold shares in AI-related businesses despite continued growth in some equipment suppliers' orders.

The important distinction is that the first four developments describe real operating needs or reported commercial activity. The fifth describes market sentiment. Falling share prices alone do not prove a bargain.

The reported Anthropic agreement works out to approximately US$2 million per megawatt of committed generation capacity. This is the price of the reported package, not Caterpillar's profit or the cost of a generator alone.

We think three opportunities could develop.

First: Generator reservation and delivery verification. Developers may pay specialists to identify equipment with genuine delivery commitments, check ownership and confirm whether those rights can legally be transferred.

Second: Gas connections, emissions permits and installation services. A generator cannot simply be placed beside a data centre and switched on. Fuel delivery, safety approval, electrical connections and testing must come first.

Third: Long-term generator maintenance and fuel-efficiency services. Once these machines begin operating, customers will need parts, inspections, repairs and monitoring. That could create recurring revenue rather than a single equipment sale.

The most directly exposed public company is Caterpillar (NYSE: CAT). On 6 November, CAT closed at US$558.63, with a reported price-to-earnings ratio of approximately 21.9. Its Power & Energy segment earned a 24.4% segment profit margin in the latest quarter, although that division includes many businesses beyond AI data centres.

For a smaller engineering business, a practical first move is to approach developers and offer a paid delivery-risk assessment. An illustrative US$40,000 contract with US$25,000 in direct staff, travel and testing expenses would leave US$15,000 before overhead and taxes. That is a hypothetical 37.5% gross margin, not an observed customer price.

What would prove us wrong? Manufacturers might shorten delivery schedules, customers might abandon gas-powered projects, governments might restrict emissions, or equipment commitments might be impossible to transfer. Most importantly, if developers will not pay for these services, the proposed business opportunity fails.

Sources and timestamps: [S1]; [S2]; [S3]; [S4].

Second to fifth consequence effect

Second-order effect: More AI companies may seek existing equipment reservations instead of joining the back of manufacturers' waiting lists.

Third-order effect: Buyers who secure generators sooner could increase demand for gas infrastructure, installation teams, electrical switchgear and local permits.

Fourth-order effect: Regions with available fuel connections and clearer permitting could attract more AI data-centre construction than regions offering land but unreliable electricity.

Fifth-order effect: Electricity access and equipment delivery rights could become important factors in determining which AI companies can serve paying customers first.

These are connected forecasts, not completed outcomes.

Analysis against the 20 intelligence layers

| Layer | Findings and limitations | | --------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------ | | 1. Money movement | Reported US$3.4 billion equipment commitment; independent transaction confirmation remains incomplete. | | 2. Early purchasing | Existing generator commitments reportedly changed hands before delivery. | | 3. Supply bottlenecks | Manufacturing and delivery schedules are central; individual Caterpillar lead times remain unverified. | | 4. Hidden beneficiaries | Qualified installation, fuel-connection, inspection and maintenance providers. No subcontractor award is confirmed. | | 5. Company exposure | CAT Power & Energy: US$8.78 billion quarterly sales, US$2.14 billion segment profit. Anthropic-specific earnings contribution unknown. | | 6. Unit economics | Illustrative service: US$40,000 revenue less US$25,000 direct costs. No paid pilot verified. | | 7. Institutional activity | Caterpillar returned approximately US$2 billion through dividends and buybacks in Q3; this is not insider proof of the thesis. | | 8. Hiring | No verified project-specific recruitment surge establishing new capacity. | | 9. Government spending | No public procurement allocation established for the reported private deal. | | 10. Foreign early warnings | No independently confirmed foreign distributor notice establishing additional generator shortages. | | 11. Consumer behaviour | End-customer purchasing is not the main driver. AI demand must ultimately convert into paid computing usage. | | 12. Market contradiction | Equipment demand is rising while AI-related share prices have faced selling pressure. Valuation explains why both can occur. | | 13. Missing activity | Actual deliveries, commissioning and customer payments are not yet established for the reported arrangement. | | 14. Commercialisation | Established gas-generator technology; the challenge is delivery and installation rather than invention. | | 15. Competitive weaknesses | Developers without reserved equipment may face slower launches, but individual competitors' positions are unknown. | | 16. Geographical spillovers | US generator commitments could affect international equipment availability; export effects remain unmeasured. | | 17. Historical OTR patterns | Related AI-electricity research exists, but no equivalent historical prediction has been outcome-audited for this specific reservation model. | | 18. Awareness and pricing | CAT's reported 21.9 earnings multiple means an unpriced stock opportunity has not been demonstrated. | | 19. Disconfirming evidence | Faster supply, project cancellations, permit failures, cheaper alternatives and already-priced profits. | | 20. Immediate trigger | Obtain a paid customer scope and authenticated equipment reservation within 30 days. Stop if transfer rights, demand or project economics fail verification. |

Research verification status: These five investigations used live public sources, original company and government disclosures, OnTheRice's published Signals and selected connected research files. Several intelligence layers remain explicitly unverified. Six independent collaborating agents were not available in this session, so unanimous six-agent approval was not achieved. The five sets should be treated as research candidates, not as five fully validated or guaranteed-profitable predictions.