Sources: NTT Data is in advanced talks to build a data center for Anthropic in Osaka of up to 500 MW, among Japan's largest, potentially costing $6.3B
THE SO WHAT
A single 500 MW, ~$6.3B facility dedicated to one AI customer in Japan shows hyperscale AI is now driving national data center footprints, not just renting capacity from them. If you’re planning serious model workloads in Asia, power, geography, and sovereign alignment just became first-order architecture decisions.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIThis ex-Pleo exec plans on taking on US rivals with his ‘AI CFO’ startup
The ‘AI CFO’ race is really a data and workflow race — whoever plugs deepest into bank feeds, ERPs, and payroll wins. If you run finance or ops, expect a wave of tools trying to intermediate your relationship with your actual CFO and decide now where you’re comfortable ceding judgment to software.
Applied AISunak, Gray, Chandrasekaran on Geopolitics, AI
When a former UK PM, Blackstone’s president, and Tata’s chair frame AI as a driver of financial flows and geopolitical realignment, they’re signaling that AI infra and data are now strategic assets. Multinationals should be treating AI footprint decisions — where you put data centers, talent, and model training — as geopolitical choices, not just cost optimizations.
Meta was prepared to write an up to $200 billion AI check before knowing whether its model would work
Being willing to contemplate a $200B AI spend before model certainty shows how far mega-caps have moved from ROI-based budgeting to strategic deterrence. Everyone else in the stack needs to assume that foundation model economics will be distorted by balance-sheet wars and design their own spend with much stricter hurdle rates.
Applied AIAnthropic says new measures like preserved thinking in Fable 5.1 and Opus 5.5 have reduced data harvesting through distillation attacks by about 90% to 99%
If Anthropic’s reported 90–99% reduction in distillation attacks holds up, model IP protection just became a lot more technically tractable. Enterprises fine-tuning or exposing APIs should be asking every vendor what their anti-distillation posture is and updating threat models accordingly.