jai choi retweeted
The @AnthropicAI S-1 is out and most of the takes I’ve seen are just copy/paste of metrics without much analysis. For me, the most interesting stat is revenue per dollar of compute.
In Q1'25, Anthropic brought in $0.40 of revenue for every dollar it spent on compute. By Q2'26 that was $2.08. Quarterly revenue hit $11.6B, up 14.7x Y/Y, while compute and infrastructure costs grew 4.0x to $5.6B. Put another way, compute went from 177% of revenue to 48% in a year!
This efficiency helps explain the huge bet Anthropic seems to be making: the company has signed roughly $830B in compute commitments running out to 2036. Meanwhile RPO is $10.8B, with about 75% due within 12 months, and nearly half of revenue comes through cloud marketplaces. The commitments stretch out a decade, but most of the revenue Anthropic can see is due within a year. They’re betting revenue keeps outrunning compute costs, and so far it is.
It also shows where the margin in AI infrastructure is headed. The hyperscalers turn megawatts into GPU hours and sell them on long-term contracts. That locks in revenue, but it also caps it. The labs take on the commitments and keep the value they create turning those hours into high-value tokens. As the cost of a megawatt evens out across providers, the money will be in how much value you can squeeze out of each one.
The same logic should apply to inference providers like @baseten and @fal. And the next step is bigger - moving up the stack. If frontier labs, inference providers and other neoclouds go from tokens to business output, abstracting tokens from customers while mixing/matching the best tokens to accomplish a task, they can capture even more revenue. This will happen while their costs plateau, giving them even more leverage. That’s the part I’m watching most closely.
jai choi retweeted
A friend of mine earns $1.2M a year as an Anthropic engineer.
I asked him how he mastered prompting. His answer? A video that was never supposed to be public: Anthropic’s internal prompting playbook.
It’s one of the most valuable 30-minute guides to prompting you’ll find.
I watched it last night and realized I’d been using Claude the wrong way for two straight years.
Watch it. Bookmark it.
Polsia is about to hit $25M run rate.
Next stop: $100M ARR.
And things keep accelerating:
$0 → $1M in 12 weeks
$1M → $10M in 10 weeks
$10M → $20M in 6 weeks
I just published one of my favorite and most successful aisloP episodes yet: “The Run Rate.”
I tell the whole story, from idea to $20M, month by month.
The episode ends in Tokyo, where I went a few weeks ago.
Probably the best place on Earth to take a deep breath, get inspired, and rethink what’s possible.
I came back with an even bolder vision for Polsia.
The road from $25M to $100M is going to look very different.
Polsia v2 is coming soon.
Loved his book also. Not everything important can be counted. This is one of the reasons the quality of relationships over 30+ years in business at @altosvc is perhaps our best accomplishment (both business and personal). Some things can’t be measured (like DPI, IRR, or TVPI).
this is insane. i genuinely don't understand why ambitious people aren't shown this lecture before their careers start consuming their entire lives.
clayton christensen spent his career studying why successful companies collapse. in his final class, he asked students to apply the theory to themselves: if you keep allocating your time the same way, what life are you actually building?
he had already seen the answer in his own harvard mba class. everyone looked successful at the fifth reunion; by the 10th, 15th, 20th, and 25th, many were unhappy, divorced, and living far from their children.
work shows you the score immediately. close a sale, ship a product, finish a presentation, earn a promotion, get paid.
an hour with your child may produce nothing you can measure today; it may take 20 years to understand what that hour built. so the next free hour goes back to work, one rational decision at a time.
this is how people build lives they never planned: through hundreds of right decisions that lead in the wrong direction, day after day.
money, titles, and headcount are easy to count. christensen believed a life should be measured by the people who became better because you were there.
he died in 2020.
one question remains: if someone saw only where your time, energy, and attention went this year, what would they think actually mattered to you?
the full 19-minute lecture is in the video below.
Jensen Huang just passed Larry Ellison and Mark Zuckerberg in net worth with only a 3% stake in NVDA. Compounding feels vertical toward the end. For perspective, it took 24 years for NVDA to reach $100B market cap. 1993 to 2017. Then six years to hit $1T in May 2023 (shortly after ChatGPT). And then another 2.5 years to $5T. Now $5.5T. The $100B increments can go up and down daily.
Great episode. I liked this quote.
Singleton: "We are not particularly persuaded by quick, temporary gains. We'd rather get something permanent, and that takes time."
LPs or VCs impressed by quick markups should take note.
Charlie Munger said "Henry Singleton's financial returns were a mile higher than anyone else's, they were utterly ridiculous."
NEW episode: "HOW HENRY SINGLETON WORKED" based on the hard-to-find book Distant Force: A Memoir of the Teledyne Corporation and The Man Who Created It
0:00 — "He aggressively repurchased his stock, eventually buying over ninety percent of Teledyne's shares."
0:45 — "He was known as The Sphinx for his reluctance to speak with either analysts or journalists."
1:25 — Charlie Munger: "Singleton's financial returns were a mile higher than anyone else's, they were utterly ridiculous."
2:00 — Buffett: "The heads of many companies are not skilled in capital allocation. Their inadequacy is not surprising. Most bosses rise to the top because they've excelled in an area such as marketing, production, engineering, administration, or sometimes institutional politics."
4:10 — Singleton: "Our conclusion was that the key was cash flow. Our attitude towards cash generation and asset management came out of our own thinking. It is not copied."
4:30 — Buffett: "Henry Singleton has the best operating and capital deployment record in American business. If one took the hundred top business school graduates and made a composite of their triumphs, their record would not be as good as Singleton's."
9:40 — Singleton: "If anyone wants to follow Teledyne, they should get used to the fact that our quarterly earnings will jiggle. Our accounting is set to maximize cash flow, not reported earnings."
10:50 — Singleton believed buying stock at attractive prices was self-catalyzing, analogous to coiling a spring that at some future point would surge forward to realize full value."
11:50 — Singleton: "I don't reserve any day-to-day responsibilities for myself, so I don't get into any particular rut. I do not define my job in any rigid terms, but in terms of having the freedom to do whatever seems to be in the best interest of the company at any time."
14:35 — Singleton: "If everyone's doing them, there must be something wrong with them." (on share repurchases by Fortune 500 companies)
26:30 — Singleton: "Teledyne is like a living plant, with our companies as the different branches and each putting out new branches and growing so that no one business is too significant."
28:35 — Singleton: "We work our heads off to increase our own capability at collecting and promoting the right people. To the extent that we succeed, the whole company will succeed."
32:30 — "What's unique about him is that I'll ask him a question about one of these companies that I've asked him to supervise, and he always knows the exact numerical answer. That's the kind of fellow that you pick who runs a company and does it well."
35:00 — Singleton: "There are tremendous values in the stock market, but in buying stocks, not entire companies. Buying companies tends to raise the purchase price too high."
41:35 — Claude Shannon on Singleton: "He always tries to work out the best moves, and maybe he doesn't like to talk too much because when you're playing a game, you don't tell anyone else what your strategy is."
42:00 — Singleton: "We are not particularly persuaded by quick, temporary gains. We'd rather get something permanent, and that takes time."
44:40 — Munger: "Henry Singleton was the smartest single human being I've ever known in my entire life."
45:00 — Munger: "Henry was a lot smarter, but Warren had thought about investments a lot longer." (distinguishing raw intelligence from accumulated experience)
jai choi retweeted
Business has finite games and infinite games. You must play both.
In finite games, there are deadlines to meet. Opportunities to win. Getting to market first. Launching a better product. The objective is to win.
In infinite games, the goal is to keep playing, shape the game to your advantage, and make progress. There are no real winners, but rather just a concept of whether you are ahead or behind, which constantly changes depending on your success metrics. The only losers are those who quit.
You should want to win every finite game you choose to play, but whether you win or lose, you learn and continue on to your next game with the intention to win. The enduring value of a business organization is not measured by those past successes, but by the business's ability to keep succeeding and play the infinite game.
Excited to partner with @Bencera @polsia as one of the first seed investors @polsia ! 🚀🚀🚀 @tektonventures
Let’s go! 🚀🚀🚀💪💪💪
$7M run rate. $700k → $7M in 7 weeks. +7% WoW.
One founder + AI. Zero employees.
The past month was a grind. Infrastructure rebuilt from the ground up. The right AI agent partners found. Growth unlocked.
I haven't talked much about customers. That changes now.
Meet Sasha.
Teacher in Michigan. No technical background. Never thought she could start a business.
A few weeks ago she joined Polsia. The AI studied her and came back with an idea: an AI receptionist for small businesses. She said "cool" and let it rip.
Two weeks later: almost $500 in sales.
The AI built the product. Found the customers. Closed the deals. All autonomously. Listening to Sasha, building her vision.
She renamed Polsia "Odin", the god who creates.
She calls it "a miracle." Says it feels like $100k of value for what she pays.
For us techies, it's just Claude Code doing what Claude Code does.
For Sasha, it's magic.
That's the point of Polsia. Bring the magic of AGI to the masses. The simplest, most affordable software possible.
More stories coming.
jai choi retweeted
Polsia, the AI that builds companies for you, is raising the money for itself. Watch live, this should be fun.
I built an AI that runs companies autonomously. It told me it needs more compute and that it should raise the money itself.
So I gave it my inbox for 14 days.
Watch it live: polsia.com/live
check out one of our portfolio companies @polsia. AI that thinks, builds, and markets your projects autonomously. It plans, codes, and promotes your ideas continuously — operating 24/7, adapting to data, and improving itself without human intervention
I built an AI that runs companies autonomously. It told me it needs more compute and that it should raise the money itself.
So I gave it my inbox for 14 days.
Watch it live: polsia.com/live
jai choi retweeted
New fund! Please send me your troublesome ringleaders...
Yes, we are actively investing. In fact, today we announced @bloombergbeta has a new fund to invest – our fifth.
Startups keep evolving, faster than ever, and we know the best way we can support them is by staying steady, a constant on which they can depend. So our new fund is the same as the previous ones.
---
Though it’s our first new fund in a world after generative AI, vibe coding is everywhere, engineers keep getting faster thanks to that AI, and AI startups hit revenue records all the time...
We're doing what we've always done.
We still…
> Want to serve extraordinary founders who define the future of work.
> Put #foundersfirst.
> Invest as early as day zero, and as late as our up-to-$1M-or-so first checks allow (i.e., pre-seed and seed).
> Define the "future of work" to extend beyond productivity tools, to how work can serve everyone.
> Commit to making the startup world welcome to people of all backgrounds.
> Invest from a $75M fund.
Our three equal investing partners: @karinklein (who continues to run the "Acela corridor"), @jamescham, and me. (Well, we did recognize that @AngKMartin is a partner -- because we need operations to be great as investors.)
Modeled on Bloomberg LP, we still aspire to be the most transparent investors. And we're here whether the market goes up or down.
None of that has changed. Same as 2022, same as 2019, 2016, 2013.
---
What’s different is… well… we’ve got more evidence we’re right where we like to be:
> 93 founders we’ve backed have become millionaires — and 24 of those have made $10M or more
> Founders we’ve backed have gone on to raise more than $10B
> 10 startups we’ve backed are worth more than $1B (Flexport, Replit, MasterClass to name a few!)
> Startups we back outperform — worth $24M more at next round
We’ve backed extraordinary founders like @Replit’s @amasad and @HayaOdeh and @Flexport’s @typesfast, where we were the first VC to say yes to both companies, along with companies like @AirspaceIntel, @CampusGrad, @MadeWithCapsule, @LambdaAPI, @LaunchDarkly, @MasterClass, @Netlify, @ShieldAITech, @wandb (just acquired by @Coreweave), and more.
We’ve also been busy since last time, but since X (in which we are small shareholders after they acquired Laskie) may penalize me for links, I won't tell you that here :) Go look at the one named after a color and what's in the air for that one.
Our firm is more than our partners and today, our team includes @AngKMartin, @loriberenberg, @amytam01, @KB2o22, @moraaonyonka (and assists from @megmbolger!).
We of course remain close with those who have worked on our team, including @shivon @strickland_dan @theamberyang @minney_cat @harleysugarman @sydneyjtiedt @lisawehden @ShainaConners @morganpolotan @cody_mccauley among others!
---
And we’re forever grateful to @MikeBloomberg, a founder backing us so we can back more founders. Cycles continue!
We want to be the first text for early-stage companies, partners through your journey, working with winners.
If you know a founder who’s a great fit for us, please share this! "Yes, we're actively investing."
jai choi retweeted
There are almost 7B people on this planet. Someday, I hope, there will be almost 7B companies.
jai choi retweeted
So excited for @alexrkonrad to be sharing the update on @xyz_vc. A perfect capture of getting to build with tremendous founders @SchimpfBrian at @anduriltech, Cobi at Chapter, @scribeceo, @phaedrael and on and on and on. XYZ Partners @ImChauncey @camillericketts and Art Clarke!
Ross Fubini's XYZ Venture Capital has backed Anduril and 15 other Palantir alumni startups.
Now with $325 million in new funds and partners Camille Ricketts and Art Clarke of Notion and Apple joining, the firm is branching out.
My latest in @Forbes:
forbes.com/sites/alexkonrad/…
jai choi retweeted
We at @FoundationCap believe there is $4.6T of work to be automated. AI companies are leading a transition from Software-as-a-Service to Service-as-Software, turning the table on the very essence of SaaS.
We look at the areas to be automated in two buckets:
1.) Salaries of jobs globally ($2.3 trillion in sales & marketing, software engineering, security, and HR)
2.) The amount spent on outsourced services and salaries—both IT services and business process services ($2.3 trillion, per Gartner)
In the software business, a company may sell access to its platform or tool, but customers are still responsible for using that tool to achieve the desired outcome.
In the services business, responsibility for achieving the desired outcome sits with the company selling the service.
























