Pinned Tweet
I started an AI company at the start of 2024 and this is how it went. 📈
Thousands of cold emails, lots of rejection, tens of software prototypes, endless AI coding, tons of learning, months living in San Francisco and an indomitable feeling that I am on the right track. 🧵
GTA VI Jozi: What if GTA was set in Johannesburg?
made by the brains behind velociraptorai.com lol
5
172
If this is real I gotta quit my job and focus on the Singularity full time geez
Holy shit. Levent Alpöge from Anthropic just claimed Claude has resolved the General Reimann Hypothesis!!!
Proof is 250,000+ pages long. Lean is released.
GRH is the 2nd of 6 Millenium prizes solved by AI, after Navier-Stokes. This may be the biggest AI math result of all time.
1
67
Ahhh I see my quarter life crisis topic of the day has arrived
the cost of choosing to work on the wrong thing is insanely high right now, higher than it has ever been in the entire history of human civilization
there are so many horse races happening all around you, it's hard to even see straight, new winners crowned every week, so many people like you must be winning big right now
and you were like, umm no thanks i really value my hobbies and quality of life, i think i'll just stay at my job, watch my skills depreciate, and hopefully retire before they lay me off
86
My appreciation for Big E grows every day - look how he's upskilling fast food workers for the AI boom!
Elon Musk reveals some of his best data center builders came from In-N-Out, because aces "making epic burgers all day long… are gonna do a damn good job of network cables"
"And then we have an in-house team, not just contractors. We have contractors too, but we have an in-house team, which we call sort of ArmyX, that is productionizing the building of the data centers."
"And doing the basic principles of productionizing things, which is where you break the tasks down into things that are easily trainable within a few days. So you don't need people that have, say, been doing networking cables for their whole life. You divide the tasks into things where you can train someone new who may have been, like, a Starbucks barista, or stocking shelves at Target. But by the way, people who are good baristas or good at stocking shelves are great workers. And also, hire someone from In-N-Out."
"Those are some of our best people. They've never done data centers before. But let me tell you, if you get some aces out of In-N-Out who have been making epic burgers all day long, perfectly, they're gonna do a damn good job of network cables."
1
102
Will post a lot of thoughts soon but a big first one so far is that if the US labs want to keep the Chinese at bay and keep their advantage they need to release some better OSS models ASAP so that people don't just move their workflows to Deepseek. commanding-heights.com/#from…
1
126
12k aircraft on contract to install Starlink = basically every palen you fly on in future will have amazing internet on it thanks to Starlink
Here are all the most important highlights from @SpaceX's Q3 2026 earnings report and earnings call:
• Total revenue: $14.58B, up 161% YoY from $5.58B and up 87% QoQ from $7.81B.
• Adjusted EBITDA: $8.81B, up 306% YoY from $2.17B and up 149% QoQ from $3.54B.
• Total Capex: $19.59B, up 185% YoY from $6.88B and 7% QoQ from $18.37B, primarily driven by SI infrastructure investments.
• Q4 Capex guidance: SpaceX expects capital expenditures of $35B–$40B in Q4 2026 as it accelerates SI compute deployment.
• Net income: $4.40B, compared to a $1.05B loss a year earlier.
• Income from operations: $2.79B, compared to a $273M loss YoY and a $143M loss in Q2.
• Q1 2027 revenue guidance: $33B–$37B in total revenue during Q1 2027.
• Annualized revenue target: SpaceX remains on track for $100B in annualized revenue by the end of 2026, based on December's revenue run rate.
• Elon's 2027 revenue projection: SpaceX could reach $500B–$1T in annual recurring revenue by the end of 2027
• Backlog: Reached $63B as of September 30, including $17.5B in deferred revenue. ~$42B is expected to be recognized within the next 12 months.
• Space revenue: $1.77B, up 6% YoY
• Connectivity revenue: $5.25B, up 69% YoY and 22% QoQ; Consumer revenue was $3.12B (+57% YoY) and Enterprise & Government revenue was $2.13B (+91% YoY).
• Connectivity operating income: $2.37B, up 99% YoY from $1.19B and 43% QoQ from $1.66B.
• Starlink subscribers: 13.6M, up 89% YoY; consumer ARPU increased to $69/month from $66 in Q2.
• Super Intelligence (SI) revenue: $7.56B, up 830% YoY and 195% QoQ
• SI operating income: Positive $431M, compared to a $1.72B loss YoY and a $1.26B loss in Q2.
• SI Adjusted EBITDA: Positive $4.95B, compared to a $428M loss a year earlier and $1.15B in Q2.
• SI Capex: $16.44B, up 245% YoY from $4.76B and 4% QoQ from $15.83B, representing ~84% of total quarterly Capex spend.
• Cloud Services Agreements: Signed additional agreements during Q3, with SI solutions and infrastructure revenue growing 192% QoQ excluding Cursor.
• SI compute economics: Newly deployed compute capacity is generating payback periods of less than 12 months at current pricing, which is slightly above the previously discussed $30–$50 per watt range.
• Cursor revenue: Contributed $608M in revenue between August 14 and September 30.
• Grok Bot: Reached 800,000 paid weekly users in early October, more than doubling over the prior month.
• Grok Bot enterprise rollout expected in a few weeks.
• Cash, cash equivalents & marketable securities: Approximately $90B at quarter-end
• Operating cash flow: Generated $13.1B during the first nine months of 2026, up 218% from $4.12B during the same period in 2025.
• Starship Flight 15: Targeted for November 2026, with SpaceX planning to attempt a tower catch of Starship's upper stage.
• National Security launches: Secured 18 additional National Security Space Launch missions.
• Starlink aviation: More than 3,800 aircraft were equipped with Starlink as of September 30.
• Starlink aviation contracts: ~12,000 aircraft are now under contract for Starlink service, representing nearly a quarter of the available commercial aircraft market.
• Starlink Mobile users: Surpassed 22M monthly users at the end of Q3, up 37% QoQ.
• Starlink Mobile U.S. launch: SpaceX plans to integrate its spectrum acquisitions and begin activating the expanded service in early 2028, with the gaol of becoming a major U.S. mobile carrier.
• SI compute: Expanded nameplate compute capacity to 1.7 GW, up from 1.4 GW in Q2 and 0.6 GW a year earlier.
• October compute target: SpaceX expects to reach nearly 2 GW of nameplate compute capacity by the end of October 2026.
• Year-end compute target: SpaceX expects 2.5–3 GW of nameplate compute capacity by the end of 2026.
• 2027 compute target: Elon expects to add at least 10 GW of incremental compute capacity in 2027, potentially bringing the total to 13 GW or more. Elon said there is a small possibility of exceeding 20+ GW of active SI compute by the end of 2027.
• GPU deployments: SpaceX is bringing approximately 200,000–220,000 NVIDIA GB300 GPUs online monthly, with plans to maintain or accelerate that pace.
• Nvidia Vera Rubin deployment: SpaceX expects its 2027 compute expansion to rely primarily on NVIDIA's next-generation Vera Rubin architecture.
• SpaceX plans to bring a training cluster with 440,000 Vera Rubin computers online by the end of Q2 2027.
• Targeting May or June 2027 for the first functional test chips from the Austin research fab.
• Terafab mass production: SpaceX is targeting Q3–Q4 2028 for potential volume chip production at its Gibbons Creek facility
1
102
The demand is quite literally insane
Elon Musk says there's no AI demand issue and explains why SpaceX rents its compute on short-term contracts: AI compute is worth more per watt every year, renewals are coming in at higher prices, and the market is a “cheese shop” with no chips left
“Our belief is that the value of compute per watt will increase over time, not decrease. And the reason for that, I think it should be obvious to people really, is that as you improve the AI software, as the models get better and better, they’re able to, you get more useful intelligence per watt. And the value of that compute is proportionate to the useful intelligence per watt. And that’s why you see things like even old computers like H100s renting for more than they rented when they were new.”
“And I think that’s likely to continue. I think we’re going to face a shortage of superintelligence compute. And that shortage is going to remain for a long time. So, I think the smart move is to do short-term contracts and to essentially place the bet on useful intelligence per watt increasing rapidly, which obviously is done, so that’s what’s happened so far.”
“And also if you look at, say, how are our short-term contracts renewing? Are they renewing at higher dollars per watt or lower? And we see them renewing at higher dollars per watt, not lower dollars per watt. We are also frankly hounded relentlessly by every hyperscaler and by dozens of smaller companies to rent more of our compute. So we could rent out our compute several times over if we said yes to everyone. That’s just the situation here.”
“That’s why I posted that sort of video meme of the Monty Python cheese shop sketch where the guy goes into a cheese shop to get some cheese and they actually have no cheese. And so he goes like, what about this kind of cheese? What about that kind of cheese? And they really just don’t have any cheese.”
“So we’ve actually tried to offload some of our inference compute to other cloud providers so that we can use more compute for training. And we get the cheese shop sketch. You know, they’ll say, like, well, we’ll ask them about what about some, you know, GB300s? Nope, don’t have any of those. What about, you know, some GB200s? Nope, don’t have any of those. What about some TPUs? Nope. What about some Trainiums? Nope. What about some, okay, how about some H100s? And like, well, we might have some overseas somewhere and then they’ll check and come back and say, nope, we don’t have that either. So it’s like ridiculous.”
“Yeah. So I don’t think we’re going to have a demand issue here. And like, it’s not impossible. I’m not saying this will happen. It’s not impossible that we will have a six-month payback period for compute next year.”
118
Generational marketing run
REMY OR LINGUINI?
for those of you who have seen ratatouille, one of the great questions in the ai age: are we remy or linguini?
are the humans in control of the combined entity (we are remy), or are the ai in control and we are simply the meat proxies (we are linguini)?
our vision with muse is undeniably the former—we are remy! we have big dreams and goals, many of which may not be appropriate for our station or our exact starting condition. but that doesn't matter—we dream like a rat who wants to be a michelin-starred chef in paris.
muse is our partner to help make that happen. ultimately, we are pulling the hairs (?), and it is our vision that ends up on the plate (?), but over time we will accomplish things that we could not do without muse.
(obviously this metaphor is super confusing because i'm saying the humans are the rat, but if you watched the movie you catch my drift)
anyway this whole mini-essay was just to justify my shitpost—pls enjoy
106
Nevermind me while I absolutely vibe to this. youtu.be/4Q-o_ylnVnc?si=PzPK…
1
163
Mitch🦖 retweeted
Announcing $38M Seed & Series A from Sequoia, First Round, and Hummingbird.
Healthleap's mission is to improve the health outcomes of as many people as possible.
Results so far:
• faster healing resulting in 1.6 shorter hospital stays per patient flagged as at risk by HealthLeap (Penn)
• 39% more appropriate diagnoses, same staff (Cedars-Sinai)
• 3 → 50+ hospitals in the past year
We're hiring across all roles.
Mitch🦖 retweeted
81% of all major math discoveries
from the last three years were released today by OpenAI. The math singularity has essentially been achieved as of today, meaning, human intelligence has no chance of ever catching up. Every scientific field will experience this eventually!



















