The Argument · Ins7ghts Daily, Thursday 8 October 2026
Every hype cycle hands me the same question, and it is the only one I trust: does this make things faster, smarter, easier, or did we just trade old tech debt for new tech debt? Yesterday handed that question a term sheet.
The evidence. A tier-one hyperscaler took 210 megawatts on a fifteen-year term, about $5.2 billion of base-term contracted revenue, for capacity with no workload on it yet. Google committed to 890 megawatts of nuclear output. An international group put nearly $2 billion into the data that disease models will eventually train on. And a money desk spent the same morning on corporate efficiency gains that are still stuck while AI spending climbs. Not one of those is a mistake. Every one of them is dated out past the point where anybody can show the return.
The best case against: lead times are driving this, not confidence. Transformers, interconnects and shells take years to arrive, so a buyer who waits for proof is not being prudent, they are standing at the back of the queue. On that reading a fifteen-year lease is not a bet on 2041, it is the price of being able to serve 2028 at all. Fair, and probably true for the people signing.
The rest of this piece is for subscribers.
Ins7ghts Daily is free. One letter a morning from Yves Mulkers on what moved in data and AI, and why it matters to your side of the table.
Already subscribed? Sign in at the top of the page.