Exponential View 🔮

Full report: exponentialview.co/p/the-state-of…

THE STATE OF THE AI ECONOMY 🧵

Once you truly understand what part of the process is increasing token consumption, you will be forced to use AI in a completely different fashion. According to the analysis presented by @ExponentialView "A coding agent that operates over 10 turns might need to re-read its full context every turn. That repetitive reading of context could use as many as 55x more tokens than a single-turn query for the same task. Actual active inference is probably only 15-20% of the total token consumption. The rest is invisible work that you, as a user, and possibly the company paying for it all, haven’t modelled." Proper documentation standard and knowing how to provide context effectively helps you reduce your token consumption. Think of token usage like fuel for your car. Don't waster it regardless of what model you are using.

We’re excited to share that AI in the Real World with Foundation Capital is now live on all major podcast platforms! Tune into our latest episode with @azeem of @exponentialview, where he and @joannezchen discuss the ways in which the great reorg is only just getting started, now on Spotify, Apple Podcasts, YouTube, and more. youtu.be/gYGyfHRwGTU

@aparanjape Want to maximize your use of AI but still give your brain enough exercise? Maybe buy a good quality notebook and fountain pen, as @ExponentialView suggests, and spend a few hours every week thinking with paper and pen. cc: @docbhooshan

Azeem and his prop research is so good. Been subscribed to substack for almost a year. Truly novel and informative

Extremely smart take on the tokenmaxxing panic and why it won’t last:

Are we nearing a compute crunch? In our latest Gradient Update, @luke__emberson and @Jsevillamol estimate how many tokens all the Blackwell chips on Earth could serve, and compare this to total token demand. Direct comparisons are difficult, but it appears demand is growing much faster than supply.

📈 Why AI bills rise as costs fall exponentialview.co/p/data-to-star…

.@ExponentialView estimates that the number of tokens processed per quarter has grown by around 17,000x over four years.

Tokens are exploding. Enterprise AI bills are getting bigger. And both trends are accelerating at the same time. 🧵

"In the US, the average monthly spend on AI by large enterprises grew 36% to $85,000 between 2024 and 2025" @azeem @ExponentialView

The cost killers are coming Via @azeem

Uber & ServiceNow blew through their 2026 token budgets in four months. AI is diffusing fast and finance is scrambling: exponentialview.co/p/monday-data-…

The U.S. is losing the AI race, because China has competition whereas we have monopolies. exponentialview.co/p/inside-chine…

It’s a fascinating period when you take a step back… a very good article on the Chinese approach. #AI #China @hannahjpetrovic @ExponentialView exponentialview.co/p/inside-chine…

A couple of weeks ago, @Bloomberg reported that TSMC won't buy ASML's newest chipmaking machine until at least 2029. ASML's High-NA EUV is the most advanced lithography tool ever built. When the world's #1 foundry passes, that's a strong signal. Today's @exponentialview analysis w @hannahjpetrovic exponentialview.co/p/the-broken-b…

Over the past week, I have been in China meeting AI and robotics teams, including Zhipu and MiniMax (the two publicly listed foundation model companies), as well as Kimi, Alibaba, Xiaomi, Bytedance, and others. Demand is booming. Zhipu, for example, is serving 5.5 trillion tokens per day. Developers are rushing to the platform, joining at about ten a minute. Keeping up with inference loads is a struggle in China, as it is in the US. Teams universally acknowledged compute constraints, particularly the shortage of Nvidia chips. But that isn’t stopping innovation under those constraints. Researchers and developers are free to use whichever model they want. Anthropic’s Claude was the preferred choice for technical teams, but it was clear that “dog-fooding” is commonplace. I'll write something in more depth in the next couple of weeks. For now, here’s a short video from one of the many demos we saw at Unitree:

Here's @ExponentialView analysis of the solar supercycle exponentialview.co/p/solar-superc…

"Golf courses in the US consume more than 30x as much water as the entire country’s data center industry uses for cooling" @azeem @ExponentialView