Advertiser Disclosure: We may earn a commission from partner links at no extra cost to you. Paid placements are labeled. Partners do not control our editorial conclusions.

Google Locks 3.6 GW as South Korea Funds a $3.5 Billion AI Model

October 6, 2026 ~2 min read
Not financial advice. All content is informational only. We may hold positions in securities mentioned. Always do your own research before making investment decisions. Affiliate Disclosure →
Full source review not recorded. Linked sources do not by themselves mean every material claim on this archived article has been reviewed.

Google secures nuclear power at grid scale while South Korea puts public capital behind a frontier AI model.


Google and Constellation assemble a 3.6-GW power package

Decoded: Alphabet's Google and Constellation Energy announced agreements covering nearly 3.6 gigawatts in the PJM market, the companies said on Oct. 6. A 20-year purchase agreement will fund 890 MW of new output from upgrades at 11 nuclear units, while a separate 15-year supply agreement covers 2,700 MW. Constellation expects more than $4.3 billion of investment, with the first upgrade due in 2028.

Why it matters: The structure shows how hyperscalers can fund incremental generation faster than a new plant while keeping existing assets economic. For Google, firm power reduces a key constraint on AI infrastructure growth; for Constellation, long contracts improve revenue visibility.


South Korea puts $3.5 billion behind a frontier AI model

Decoded: South Korea plans to begin a $3.5 billion program next year to develop a frontier AI model, Reuters reported on Oct. 6. The commitment moves the model race beyond corporate balance sheets and into national industrial policy.

Why it matters: Public funding can deepen demand for compute, data infrastructure and AI talent, but the investment case depends on whether the program produces commercial deployments rather than a research-only model. It also adds another sovereign competitor to a field led by U.S. and Chinese labs.


Stay decoded. See you tomorrow.

— The Get AI Decoded Team