Seeking Win-Win-Wins For fun: https://xitter.cf/t.co/NSNAbUz0Rg, a game for Survivor superfans
Austin, TX
Joined July 2018
- Tweets6.4K
- Following1.1K
- Followers511
- Likes11.9K
Pinned Tweet
What is AI's ROI?
That's the quadrillion-dollar question.
But it's more like:
* What was AI's ROI in 2023? (small)
* What was AI's ROI in 2024? (medium)
* What is AI's ROI in 2025? (large)
And:
* Stock analysts already know our personal productivities have increased by a lot.
* Software companies already know the cost to create new code and content is being decimated (See: Duolingo).
* CEOs are just now starting to realize the huge potential for better analyzing their businesses using AI
* Teachers are just now realizing the huge potential
This general unlock (🫡 General Unlock) is happening person by person. We're all getting a very good personal assistant (former price: $100,000/yr) for $20/month.
Before, productivity was mostly a function of time and energy. Now productivity is a function of time, energy AND MONEY. And there's no shortage of money.
In fact there's a technological revolution in money. New monies are being created every year. Often better money. Or did you think gov't-issued fiat money or gold would be the end of money's evolution?
Isn't the productivity costly in energy units though?
Luckily we're experiencing non-stop improvements in energy production. Fracking efficiencies continue in natural gas. PV cells are constantly improving. And solar storage (via lithium batteries) is a game-changer.
But how much of the world is getting this huge productivity boost?
Wait-- only 11 million ppl pay for ChatGPT? That's bonkers. It's easily the best productivity tool in history. (Apple MacBook now takes 2nd place for me.) This number will surely keep skyrocketing.
And Starlink is providing fast internet to the half of the world that doesn't yet have it. Fast internet = AI productivity boost.
But what if AI doesn't continue to get better like it has?
Even if AI doesn't significantly improve in 2026, the current versions of AI will drastically improve productivity for more and more people. But I wouldn't bet against AI improving significantly.
Who will harness and channel and organize these massive gains in productivity?
Luckily we're also experiencing a surge in founder skill and courage. Business leaders are following Elon Musk and others and doing the hard things.
And then comes the next-generation of AI-native founders, aka The Wizards, who will only know a world where creation is the norm....
Tim McCollough retweeted
6 things I’m watching for $SPCX SpaceX Q3 earnings tomorrow:
1. @Grok @bot adoption in Q3 and management’s Q4 forecast.
2. Whether the $100B year-end ARR goal gets reaffirmed or raised above it.
3. Progress toward 2+ GW of terrestrial compute by December, and the 2027 outlook versus Elon’s possible 10 GW.
4. If 10 GW is achieved next year, how much is internal use versus leased out.
5. Starship development updates.
6. @Starlink 🌐 subscriber growth rate.
Musk's choice of steel for Starship looking better and better.
Rocket Lab had an engineering hiccup because of carbon fiber. (Problem solved though, they said.)
Today we shared an update on Neutron's development, including adjusting the schedule for flight one to 2027.
Full details: investors.rocketlabcorp.com/…
Reminds me of how "Information Superhighway" (1994-1995) was supplanted by "the Internet."
I liked "Actionable Intelligence," but Super Intelligence is just better branding.
"The walkie-talkie spectrum from 2004 ended up being the piece that lets satellites replace cell towers. Nobody who touched it for twenty years saw that coming."
$ECHO's Charlie Ergen saw it coming.
T-Mobile spent five years trying to get rid of the exact spectrum Elon just called "the last critical piece" of complete phone coverage in America.
This 800 MHz block is Sprint's old Nextel spectrum, the stuff that powered push-to-talk walkie-talkie phones in 2004. The DOJ forced T-Mobile to divest it as a condition of the $26 billion Sprint merger. Dish had first rights at $3.59 billion and walked away because it couldn't raise the money. T-Mobile finally offloaded it to an investment firm called Grain Management last year for $2.9 billion plus a spectrum swap.
Grain's original plan was selling it to utilities for private networks.
Fourteen months later, SpaceX buys the whole portfolio, and shares of the carriers who passed on it fall as much as 8% in a single afternoon.
Here's why it matters so much. Starlink's satellite signal at 2 GHz can reach your phone outdoors, but it dies at the first wall. 800 MHz penetrates buildings. Most phones already have the band built in. So satellites handle the open sky, this low-band layer handles indoors, and suddenly Starlink Mobile covers everywhere a normal carrier does without the tower network.
Stack the moves. $17 billion for EchoStar's 50 MHz of 2 GHz spectrum. FCC approval for 15,000 Gen2 satellites this week. Now the low-band coverage layer.
Verizon, AT&T, and T-Mobile paid $81 billion in a single auction for C-band spectrum in 2021, then spent years and tens of billions more deploying it on towers. SpaceX assembled a nationwide carrier out of a distressed-asset sale, a forced divestiture castoff, and rockets it already owns.
The walkie-talkie spectrum from 2004 ended up being the piece that lets satellites replace cell towers. Nobody who touched it for twenty years saw that coming.
"Spectrum wars"
$ECHO
The invisible hand sends scarce resources to whoever uses them to create the most value, fastest.
SpaceX has shown it's the best at producing AI.
So $SPCX will get all the compute, power, and capital it can handle; and the market should expect its AI production to keep ramping
Elon’s latest GPU schedule here implies ~2.7 GW of server power (~3+ GW facility) by year-end if the December tranche arrives on time & online vs. his comments on SpaceX’s August expectation of “over 2 GW" back in August.
The math, using its filing’s initial cluster figures:
>GB200: 210 MW ÷ 110k GPUs = 1.91 kW/GPU
>GB300: 220 MW ÷ 110k GPUs = 2.00 kW/GPU
>December: 110k GB200 × 1.91 kW + 1.10M GB300 × 2.00 kW + ~300 MW estimated for Colossus 1 = ~2.7 GW
That puts combined server power at ~1.4 GW now, ~1.8 GW next week and ~2.3 GW in November. +Epoch’s 14% allowance for other IT equipment takes December to ~3.1 GW IT, before cooling and facility losses.
This schedule suggests upside to the earlier outlooks.
The "fix" is trust.
Trust has always mattered but it's never mattered more.
In 1970, an economist wrote 13 pages explaining why good used cars vanish from the market. Three journals rejected the paper as trivial. It won the Nobel Prize, and it just predicted the death of the compliment.
George Akerlof's insight was simple. Sellers know if a car is a lemon. Buyers don't. So buyers only ever offer the average price, anyone with a genuinely good car refuses to sell at that price and walks, and the market slowly fills with junk until buyers lower their offers again. The spiral runs until only lemons trade.
The engine of the whole collapse is unverifiable quality.
Now look at what "is this AI?" actually is. Praise was the currency creators got paid in. "So well written" really meant "I can see the hours you spent." The compliment was payment for verified effort.
AI made effort unverifiable. A stunning illustration might represent 40 hours or 40 seconds, and you can't tell which from your phone. So the rational reader does exactly what Akerlof's car buyer does. He stops paying full price to anyone and pays the average instead, and the average assumption online in 2026 is "probably machine."
Then the spiral kicks in. Human artists bear real costs, so lemon prices push them out first, which raises the share of generated content, which makes the skepticism more accurate, which drives out more humans.
Akerlof needed used cars to prove the theorem. We're running it on everything people make.
His paper's fix wasn't better cars either. Warranties, brands, dealer reputations, mechanisms that make quality verifiable again. Which is why the next decade of the creator economy will be a war over proof of process rather than proof of talent.
Tesla Semi Huge Price Increase
Ryder had secured $7.5 million in funding to buy Tesla Semis. At $180,000, this would have bought 42 Semis. But in a new letter, Ryder says they are now only able to buy 18 Semis with that same $7.5 million.
Ryder mentioned "dramatic changes to product economics." 😆😆 That's one way to put it!
Conclusion: Tesla Semi is now being priced over $400,000
That price tag would make sense if Tesla Semi actually delivers over $60,000 in savings per year.
If those savings are real, then even at $400,000 there will be excess demand and a long waitlist to get a Semi off the production line.
As I wrote last year, the market is/was undervaluing Semi 👇
Trust premium 📈 📈
Internet supercharged distribution.
AI is supercharging production.
So many great products. Which will you choose?
Who do you trust?
Gates → Zuckerberg → Altman
I spent much of the weekend talking with the team who did this work. Seb--and everyone else--acted with integrity and generosity throughout.
Initially we believed the other team had also solved the problem. We wanted to collaborate and do a joint release.
When we learned that they had Euler but not Navier-Stokes, we offered to let them go first, to suggest that they should be the ones to get the prize, and optionally for Tristan to be the lead author on a rewrite of the OpenAI proof. We felt it was challenging to offer the same to Levent (an Anthropic employee), who was not willing to talk or coordinate with us anyway. We were open to other solutions.
We would have greatly preferred coordination. We did not rush to publish even though the other team wasn't communicating with us. The team threatened us with unfounded accusations of plagarism.
Now that we can see their work, the approaches appear to be different. It is also worth noting that our latest model can solve many, many other math problems.
It is true that we tried this because there were rumors on the internet last week that Anthropic's models had solved a millennium problem and we were curious if ours could do it too.
OpenAI can't be trusted not to use the insights you produce on ChatGPT or Codex?
>math professor spends a year on one of the hardest unsolved problems in math
>"Navier-Stokes"
>his drafts for the whole project went through codex sessions
>openai had his logs in codex
>suddenly rumor leaks AI solved it
>he emails openai to ask what's going on
>openai calls within days
>"funny story, our model also solved it"
>using his exact approach
>asks openai: did you train on my sessions?
>no answer
When advertising pills, companies are forced to talk about side effects by law.
Talking about the companies for which you're invested should also come with a simple line, by law: "I've invested a bunch of money in this company."
We shouldn't have to guess at incentives.
Last time $SNOW hit $375 in 2021 they had $1 B in revenue. Today its $6 B with accelerating 37% revenue growth & expanding operating margins driven by AI use cases like CoCo & Cowork. Congrats. 💪💰📈
probably nothing
Another record quarter! 🚀
Record Revenue: $47B ⬆️58% Y/Y
Record ISG Revenue: $31.8B ⬆️89% Y/Y
Record AI Servers Revenue: $16.4B ⬆️100% Y/Y
Record Servers & Networking Revenue: $10.5B ⬆️122% Y/Y
Record Q2 Storage Revenue: $4.9B ⬆️26% Y/Y
Record ISG Op Inc: $4.8B ⬆️225% Y/Y
Record AI Orders: $60.9B
Record AI Backlog: $95B
Record Commercial Client Revenue: $13.2B ⬆️22% Y/Y
Record Diluted EPS: $6.34 ⬆️273% Y/Y
Record non-GAAP Diluted EPS: $7.04 ⬆️203% Y/Y
During the quarter, we returned a record $4.3B to shareholders through share repurchases and dividends.
Demand exceeds supply, and we are still supply-constrained.
Grateful to our customers, partners, and the @Dell team who make it all possible. Onward!🙏🚀
#PlayNiceButWin