Meta projected spending up to $10bn a year on Anthropic’s AI models
THE SO WHAT
Meta internally modeling up to $10B a year on Anthropic’s models shows that even hyperscalers are willing to be major customers of rival labs when capability and time-to-market justify it. For enterprises, this reinforces a multi-model future—your stack should assume you’ll mix and match providers, not bet everything on a single house model.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIMore Evidence Emerges That Gen-Xers Feel the Most ‘Whatever’ About AI
A Gen X cohort that is neither hyped nor hostile to AI means a large slice of mid-career managers may quietly underinvest in adoption. If your change program runs through this layer, you’ll need explicit incentives and proof of value—not just corporate slogans—to move behavior.
Applied AIShadow AI is a security problem, but the EU AI Act makes it a legal one
Once the EU AI Act lands, employees pasting corporate data into unvetted tools stops being just a CISO headache and becomes a compliance and liability issue. This week, you should be inventorying AI usage and locking in sanctioned tools—after-the-fact policies won’t help when regulators ask where the data went.
Applied AIThe AI layoff narrative is missing the real story
If fintechs are blaming AI for layoffs while the underlying drivers are weak returns and cost pressure, AI is being used as narrative cover for classic restructuring. For operators, that means AI budgets will increasingly be tied to hard P&L moves—be ready to prove impact or be folded into headcount stories.
Applied AIEurope has ordered a €388M AI supercomputer from its own state-owned champion
Europe putting €387.8M into the LUMI-AI supercomputer built by state-owned Bull is about sovereignty as much as speed—compute is being nationalized as strategic infrastructure. If you’re building on European AI, expect more incentives to use EuroHPC resources and more scrutiny on routing workloads to non-EU clouds.