01 / ai economics · 11 records
What AI costs, how it makes money, and where the numbers stop matching the narrative.
AI-native and AI-enabled software companies differ on cost attribution, spend predictability and gross margin, even controlling for size. For investors, the question is whether AI was added to the product or the business was built around its economics.
rec 01.010
96% of CFOs in Cloud Capital's research said multiple AI cost categories exceeded forecast in H1 2026. The largest overruns were in compute infrastructure and data, ahead of model and token costs.
rec 01.009
Models will improve and tokens may get cheaper, but enterprise technology history suggests value accrues to whoever becomes the default, trusted choice.
rec 01.008
Gold has LBMA, oil has Brent, natural gas has Henry Hub. Compute, now central to AI unit economics, still has no reference price.
rec 01.007
volatility > performance
rec 01.006
As growth normalises, the questions become who absorbs volatility and where the economics settle, as happened in payments infrastructure. Scepticism here is about value capture, not the technology.
rec 01.005
Hyperscaler CapEx does not stay with them. It always rolls downhill.
rec 01.004
This is the gravity moment for GenAI: when the technology stops selling itself, and the economics of how you package, price, and deliver it start deciding the winners.
rec 01.003
Why AWS’s quiet policy shift this summer could accelerate the reckoning for AI startups still running on imaginary margins.
rec 01.002
AI startups are finding out the hard way: AI is spenny.
rec 01.001
The media calling Lovable a “vibe coding unicorn” is proper misleading. It might describe who they serve. It does not describe how they got there.
