📈 Data to start your week
Exponential View Azeem Azhar
AI models are shipping faster, and old ones are losing value quickly. Nvidia still doubled revenue, but new data centers barely move local power bills.
Based on reporting by Exponential View, Azeem Azhar — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Exponential View’s latest weekly data roundup points to a simple pattern: the AI economy is still moving fast, and the economics are getting harsher for anyone trying to sell yesterday’s model as if it were new.
The group says more models have shipped since its State of the AI Economy report in June. That earlier report argued that frontier models are a rapidly depreciating asset, and it says that still holds even at the highest GPQA Diamond grades, where the pricing power of a fresh model fades quickly.
The demand side has not cooled off. Nvidia said revenue more than doubled year-on-year to $96.2 billion last quarter. Exponential View also says frontier model capability has accelerated since April 2024, which helps explain why the top end keeps getting reset.
There is one more signal in the mix, and it cuts against some of the usual panic about AI infrastructure. On average, hyperscale data centers in the US appear to have only a marginal impact on nearby households’ power bills. So the market is still paying up for AI, while the local electricity bill horror story looks a lot smaller than the headlines suggest.
My take — AI-written commentary, not fact-checked reporting
The real story here is not that AI is booming; it’s that the moat keeps shrinking. Frontier models are looking less like prized assets and more like melting ice cubes with a training bill attached. That is exactly the kind of market that rewards speed, punishes complacency, and makes yesterday’s bragging rights look a bit embarrassing.
Read more about this at: Exponential View