rubicon59

Jefferies moving forecasts by GW over the course of weeks! "Satellite imaging data now points to ~27 GW of data centers by '27 (up from low-20s just weeks ago) and we see SPCX as single-handedly skewing estimates higher." GW seems like a tech spec vs massive infrastructure project but for example, going from 20-> 27GW: -+$350BN capex -+$560BN monetized inference opportunity (illustrative) -7 San Franciscos worth of power It's a crazy time cause sellside and buyside treat these inputs very flippantly when in reality they are MASSIVE What is clear, though, is the project-level trackers (e.g., SemiAnalysis, Aterio, etc.) increasingly seeing huge lines of sight. Few months ago I would've told you impossible to get ~30GW/yr but it seems like construction trends point to that being true (on a facility basis) for 2027 and 2028. Just because it's under construction and scheduled to come online mean it hits the target, but clear the ambitions and tracking of real construction progress continue to move up... *Not going to dive into the separate debate about what level of GW is needed, etc.

Gavin Baker says the Anthropic S-1 will break a lot of investors' brains "The Anthropic S-1 is going to be really important. And it's going to break a lot of people's brains. There are a lot of macro and value investors who are very confidently making these prognostications about AI with the assumption that tokens are subsidized." "Anthropic is generating cash. They're profitable. Open source tokens are profitable. OpenAI, if they're not generating cash, they will be imminently. SpaceX, same thing." "These macro and value investors are ignorant. They're very smart, but they're ignorant of the facts. They're making an assumption that tokens are subsidized and this is all going to collapse in some circular financing bonfire." "They're just wrong. The overwhelming majority of tokens are profitable for everyone in the chain. Everyone." "It's saying I'm very bearish on the world economy because oil is at $500 a barrel. If oil was at $500 a barrel, that'd be a good reason to be bearish. It's just not. They're just wrong. The central fact is wrong."

Silly analysis. Robot RAM needs will exceed everything else on Earth, as you need the robots to be on physical substates in order to shape atoms. Super intelligence needs will vastly exceed that, but ~100% of SI will be deep space compute. Obviously.

Elon Musk now literally holds the off switch on Anthropic's $2 trillion IPO. Anthropic's filing shows it has agreed to pay Elon's AI company up to $120.5 billion for computing power. And starting early next year, either side can walk away with 90 days' notice. The deal is with SpaceXAI, the SpaceX unit that used to be xAI, and it rents Anthropic the Nvidia chips that run Claude. In May, the price was $1.25 billion a month for one data center in Memphis. Reuters reported the confidential draft of this filing put the total at $84.5 billion. The public version filed today says $120.5 billion. Once all the capacity arrives in December, the bill jumps: Anthropic expects to pay Musk about $3.1 billion every month. Anthropic reportedly brought in $16.8 billion last quarter, which works out to about $5.6 billion a month. So more than HALF of what Anthropic earns today is about to go to one supplier. Now here's where it gets really insane: The lock-in on those contracts ends in January and March 2027. After that, either side can cancel with 90 days' notice. Fortune reported in July that the clause gives SpaceX a way to take its chips back if Grok suddenly needs more. Grok competes directly with Claude. And in February, Elon publicly called Anthropic's AI evil. So the company selling itself to the public as the safe AI lab is about to run a huge share of its product on hardware owned by its loudest critic. That critic also owns a direct competitor, and he can start the 90-day clock in early 2027. The filing does warn that some of Anthropic's compute suppliers build their own AI models, and that warning reads very differently once the supplier is Elon. And Elon needs this deal just as badly: SpaceX's CFO has promised $100 billion in annual revenue by the end of this year. At full speed, Anthropic's rent comes to about $9.3 billion a quarter. That's more than the roughly $7.6 billion SpaceX's entire AI business brought in last quarter. So 2 of the biggest IPOs in history are now holding each other up, and both of them have a 90-day exit. Anthropic has a real argument here. Computing power is the scarcest thing in AI right now, and Amazon's cloud capacity for 2027 is reportedly largely sold out. The filing even admits Anthropic badly underestimated how much compute it needed in the past. Grabbing every chip on the market is the rational move, and the same clause lets Anthropic leave if cheaper capacity shows up. But public buyers deserve 2 things before the listing: Anthropic should publish how much of Claude runs on SpaceX hardware and what happens to its customers if that notice ever arrives. And it should get the 90-day exit removed for the full length of the contract, the same way its other big compute bills are locked in. Until then, anyone buying ANT at $2 trillion has to answer one question for themselves: Who do they trust to keep Claude running in 2027, Anthropic or Elon Musk?

SpaceX's Raptor 3 is taking rocket engineering to an entirely different level Compared with the legendary RS-25, Raptor 3 delivers 34% more thrust while weighing 57% less, giving it over 3× the thrust-to-weight ratio To put it simply, SpaceX built an engine that's smaller, weighs less than half as much, and yet packs even more thrust than NASA's much larger RS-25. It's really hard to comprehend how much engineering went into achieving this And what's even more impressive is that SpaceX achieved all of this with a radically simplified engine designed for mass production and rapid reuse This is the engineering revolution powering Starship

The revenue side of that efficiency jump, straight from the S-1's quarterly table: $376M to $11.6B in five quarters, and Q2 flipped to +$974M operating income.

Acabo de ver cómo SpaceX acaba de confirmar que van a llegar a 13 gigavatios de potencia de cómputo el año que viene, y que hay camino claro hasta los 23 GW. Con los precios que están cobrando a los clientes, eso se traduce en ingresos anuales de entre 500.000 millones y 1 billón de dólares. Lo dijeron tanto Elon como el CFO Brett Johnson en la call. Eso convertiría a SpaceX en la empresa más grande del planeta el año que viene, no “algún día”. Están instalando 200.000 chips GB300 cada mes. Lo que antes les costaba un año entero ahora lo hacen mensualmente. Eso representa el 25 % de toda la nueva capacidad de cómputo que se va a instalar en Estados Unidos el próximo año. Cada mes puede sumar entre 30.000 y 50.000 millones de dólares de ingresos anuales. Estamos entrando de lleno en la era de la superinteligencia. Las instituciones se acaban de enterar de que esto es real. Seis bancos subieron sus precios objetivo en un solo día: JPMorgan de 240 a 500 dólares, Barclays de 254 a 300, Wells Fargo a 274, Citi a 240, UBS a 230 y Bank of America a 275. Adam Jonas de Morgan Stanley lo dejó claro: que otros te paguen la infraestructura de cómputo que luego vas a usar para entrenar tus robots es un plan excelente. Elon está gestionando personalmente la construcción del data center en Memphis (cuatro días a la semana) y SpaceX está pagando un 20 % extra a los empleados que se ofrezcan voluntarios para montarlo. El capex de los próximos cinco trimestres puede situarse entre 40.000 y 70.000 millones de dólares solo para llegar a esos 13 GW. El gran beneficiado es Nvidia, que se lleva entre 83 y 89 céntimos de cada dólar. Los 13 GW completos generarían unos 414.000 millones de ingresos para Nvidia de un solo cliente. El año pasado Nvidia facturó 216.000 millones en total. El coste de la infraestructura se recupera aproximadamente en un año de ingresos (y podría bajar a solo seis meses). Hay camino claro a duplicar las cifras y llegar al billón. La estrategia del “neo cloud” consiste en usar las ventajas de SpaceX en inteligencia por vatio, por dólar y por segundo para financiar la expansión. La inteligencia física tiene un efecto multiplicador enorme. En Starlink también hay novedades fuertes. La FCC acaba de aprobar espectro adicional para la red móvil satelital de próxima generación, convirtiéndola en la primera que combina espectro satelital y terrestre. El CEO de United confirmó que los clientes ya están cambiando de aerolínea por el WiFi de Starlink (especialmente desde Delta). Ya tienen más del 50 % de los vuelos equipados y terminarán la instalación el año que viene. El programa de status matching se multiplicó por cuatro y una cuarta parte de esos clientes pidió tarjeta de crédito de United. SpaceX ya no es solo una empresa de cohetes. Es una compañía de superinteligencia, de telecomunicaciones globales y de infraestructura dominante. Los chips ya no son el cuello de botella; lo son la energía y la infraestructura, y nadie construye eso mejor que Elon. Los números de la call superaron casi todos los modelos previos, sobre todo en data centers. Hay terreno en Texas y Memphis donde se están construyendo varios mini data centers, y cada uno puede valer del orden de los 100.000 millones. Los lanzamientos de Starship van camino de ser semanales en la primera mitad de 2026 y dos veces por semana en la segunda mitad, y probablemente superen esas cifras. Si esto se cumple aunque sea a medias, SpaceX se come el mercado de cómputo, financia la inteligencia física con el dinero de otros y se convierte en la empresa más grande del planeta mientras Nvidia se llena los bolsillos de paso.

Small piece of Lip Bu's CASPA chat last night 'Lip Bu we are going to go deep with you because we trust you.' youtu.be/aT5Z9FAmA_0?si=YG2K… Posted by Token Per Watt $INTC

SpaceX’s $SPCX Starlink just surpassed AT&T $T Fiber in subscribers after tripling its customer base in just seven quarters. What’s even more impressive is that Starlink is adding roughly 1.6 million subscribers per quarter versus about 400,000 for AT&T Fiber — showing how much faster satellite networks can scale compared to laying fiber street by street. Starlink’s connectivity CAGR sits at over 90% while AT&T Fiber manages around 17%. In a world racing toward AI / SI and orbital infrastructure, that gap isn’t just a statistic. It’s a structural advantage. Image data courtesy of Fiscal.ai

Agentic demand is growing faster than I’ve ever seen anything grow

The unthinkable is about to happen to SpaceX