Head of Strategy and Ops @MassedCompute Formerly: @bcg, @ripple
Joined January 2022
- Tweets39
- Following78
- Followers12
- Likes151
Pinned Tweet
I just dropped part 1 of my deep dive on the frontier model labs. Take a look if you want a head start on how to think about this layer of the AI stack as Anthropic and OpenAI gear up for their IPOs tinyurl.com/2wk8xs7u
Josh Klein retweeted
Today, we’re launching Finn, the world’s first AI Wealth Manager.
Managing money is exhausting. You have to figure out taxes, invest your cash, plan for retirement, and somehow know when something important changes. And even when you know what to do, you still have to go do it. Not once. Over and over again.
Not anymore. Meet Finn. It knows your money, plans ahead, and gets things done.
Finn can:
- Allocate your cash smartly
- Rebalance your portfolio
- Plan your next home purchase
- Find opportunities to save on taxes
…and so much more!
It’s like having a personal banker, investment advisor, and tax expert in your pocket.
First 50 people to comment and save this post gets free access to Finn for 3 months!
Josh Klein retweeted
“Yeahh I dumped my $NVDA to go buy the next bottleneck in the supply chain. Ticker symbol is $0028.0006 it trades on the Sri Lankan stock exchange. They make some kind of glue or something man i don’t even know haha, but it’s a big bottleneck i heard”
It's going to be interesting to see how companies test candidates' ability to use AI efficiently and effectively. I feel like somebody has to have a startup working on this…
Josh Klein retweeted
What’s happened is that we went from AI chat tools that were relatively cheap and had small context windows, to AI agents that have giant context windows, the ability to keep track of longer running work, and models that cost an order of magnitude more on inference because they’re that much better.
This has compounded far faster than most realized (unless you were paying close attention at the middle or end of last year, which many here were), and the dollars flowing in now are much more real.
What follows is a continued march of AI capability that will continue to be used by anyone with a frontier use-case (like coding, sciences, finance, consulting) and then a peeling off of tasks to lower cost models that are capable enough for the job. Whereas we thought the cost of AI might converge on a single low price per token before, it’s clear the stratification is only widening based on the task you need performed.
This will be yet another component that has to be figured out for broad AI diffusion. Enterprises will need to put in programs, new finance teams, and technology solutions to manage this all. The labs and platforms that can ensure customers can price optimize for the task at hand will be in the best position.
🦔Microsoft canceled its internal Claude Code licenses this week after token-based billing made the cost untenable, even for a company with effectively infinite cloud resources. Uber's CTO sent an internal memo warning the company burned through its entire 2026 AI budget in just four months. American AI software prices have jumped 20% to 37%, and GitHub (owned by Microsoft) is dropping flat-rate plans for usage-based billing across its products.
My Take
The AI subsidy era is ending in real time. The same company that put $13 billion into OpenAI and built the Azure infrastructure powering most of Anthropic's compute just looked at the bill from a competitor's coding tool and decided it was not worth paying. That is not a productivity failure on Anthropic's end. Token-based pricing is forcing every enterprise customer to confront the actual cost of running these models at scale, and the number turns out to be far higher than the flat-rate experiments suggested.
This ties directly to my Gemini Flash post yesterday. Anthropic, OpenAI, and Google all raised effective prices in the last six months. Enterprises that built workflows assuming AI costs would keep falling are now watching annual budgets evaporate in months. Two outcomes look likely from here. Either enterprises scale back AI usage to fit budgets, which slows the revenue ramp the labs need to justify their valuations ahead of IPOs, or the labs cut prices and absorb the losses, which makes the unit economics worse at exactly the wrong moment. Both paths land in the same place, the numbers stop working, and somebody has to take the writedown.
Hedgie🤗
$NVDA's earning just now. Exceptional quarter and good signal for broader AI industry.
Here are my initial top takeaways heading into the call:
1. Demand is very strong, evidenced by a comfortable beat on top line estimates ($81.6B vs. $79B consensus) and great outlook for next quarter
2. Progress diversifying its revenue base from hyperscalers - this is a master maneuver and a strong decision to segment out.ACIE (AI Clouds, Industrial & Enterprise) from hyperscalers so the market can track progress as the concentration risk eases. Look for commentary here on the earning call
3. Record guidance assumes zero Hopper shipments to China. With that market still locked and that baked into guidance, any movement on that front is pure upside
4. They're printing cash ($50.3B operating CF) and returning capital to shareholders via a dividend hike ($0.01 -> $0.25 per share) and an $80B buyback.
Investing in OpenAI and Anthropic near ~$1 trillion is a bet they can figure out a durable moat (a real problem they haven't yet cracked). Today, frontier LLMs behave less like software and more like depreciating inventory, and switching costs are effectively zero
Josh Klein retweeted
every layer of the ai stack is moving so fast that the economics are the only thing slow enough to study.
Some $AMZN earnings takeaways (I care mostly about AI)
1. AWS re acceleration. +28% YoY, or fastest in 15 quarters
2. Bedrock growth as a proxy for enterprise AI adoption. 170% QoQ growth in customer spend and tokens in Q1 > all prior years combined
3. Capex confirms cycle. +$59B YoY infra spend; FCF near zero → we're still in the early innings of multi-year AI buildout
Some $GOOGL Q1 takeaways:
1. Cloud +63% YoY → fastest among hyperscalers
16B+ tokens/min (+60% QoQ)
2. Gemini gaining (MAUs +40% QoQ)
3. Margins expanding (rare in AI)
4. Capex +107% → offense on AI infrastructure
Josh Klein retweeted
One of the most substantive classes with @ChaseLochmiller at Stanford. We went deep on economics of the datacenter:
- Where is the ~$650B of AI infra capex actually going this year?
- Who's capturing the margin, who's getting squeezed?
- How the bottleneck has moved from GPUs to power, and where it goes next
- The economics of neoclouds
Josh Klein retweeted
Introducing Claude Opus 4.7, our most capable Opus model yet.
It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back.
You can hand off your hardest work with less supervision.
ALT Claude Opus 4.7 Benchmarks
For years, VCs heavily subsidized Uber and Lyft fares. In ~2016, there was actually a product called Uber Pass where it cost $5 to Uber anywhere in Seattle city limits.
This is where we're at in AI. Enjoy the free credits while they last. When the VC subsidy burns off, inference pricing will look a lot more like what it actually costs to run a 50MW data center.