Latest update · 2026-10-02

Daily data refresh: Tencent's $7B chip lease, Amazon's 20-year nuclear PPA, the halted PJM auction, TVA's data-center tariff, California's AI-layoff disclosure law

Today's AI-economy briefing, verified and folded into the data. Capex: Tencent signed a ~$7B five-year lease for ~100,000 advanced AI chips from Oracle data centers in Southeast Asia — the largest overseas lease on record, and a live demonstration of how US export controls govern physical transfer but not remote compute use. Japan's JERA, Dell, and RHAELM signed an MoU for a >$15B, 400 MW Chiba AI data center (behind-the-meter, skipping the grid queue); the FT estimates the broader Japan push at $140B. The Brookings BPEA draft now gives a unit cost for the boom: ~$8.2B per 200 MW campus (~$41B per GW), with a warning that financing risk is migrating to opaque off-balance-sheet structures. Energy: Amazon and Constellation signed a 20-year nuclear PPA covering 690 MW at Calvert Cliffs (~190 MW of new capacity, >$3B investment) — the grid's biggest customer becoming its builder. PJM's emergency 6.8 GW backstop auction was halted outright after FERC's cost-allocation objections (effective date pushed to Feb 28, 2027; TD Cowen warns of 'acute uncertainty'). And TVA's data-center tariff took effect: a dedicated rate class with a Capacity Commitment Charge, making AI loads pay for their own grid. Labor & policy: California's SB 951 requires employers to disclose AI-caused mass layoffs under the state WARN Act — the first law turning AI displacement into a reporting metric. Perspectives: Goldman's new agentic-commerce note maps a $2.6T US consumer opportunity as AI agents take over shopping, with payment networks as the winners (Mixed). Anthropic's IPO prospectus warns government attitudes could hurt its commercial ties — and that advanced AI could pose 'catastrophic or existential risks to humanity.'

See the full changelog
SundayPyjamas Impact Foundation

Impact Suite

Mapping the AI buildout as a market — who builds, who funds, who pays, and what changes for people over the next decade.

New here? Start here — pick your 10-minute path.

Live impact mapNew · core update

The AI data center buildout: the power edge

What the buildout is actually doing to power — traced three orders deep, every figure sourced. Not a pitch; the data.

Global data-center electricity use (TWh/yr)

Surged ~17% in 2025 to ~485 TWh; set to roughly double to ~950 TWh by 2030. [11] IEA [1] IEA

This electricity has to be paid for — the next chart shows where the money comes from.

Big Tech capital spending ($B/yr)

The five largest tech firms spent $400B+ in 2025 and are set to raise it ~75% in 2026 — most of it AI infrastructure. [11] IEA

This spending builds the data centers that consume the electricity above — and drives the price spike below.

Cost of reserving future electricity — eastern US grid ($/MW-day)

From $28.92 to the $333.44 cap in three years; the grid's market watchdog pinned ~63% of the jump on data centers. [12] RTO Insider [5] Utility Dive

This cost gets passed to electricity bills — that's where regular people start paying for the AI buildout.

Jobs per ~100 MW data center campus

Thousands to build (~800–1,200 at peak); ~25–50 to run it. Billions of capex, dozens of permanent staff. [8] Brookings

The construction boom is real but temporary — and it pulls workers from housing and other projects that also need them.

Explore the full map

Every figure sourced · who wins & who pays, three orders deep · as of September 2026