If you run the numbers on Cybercab, the potential really is insane. Fare revenue could be bigger than all their car sales combined. Maybe even bigger than Amazon. First, for a city like Austin, how many Cybercabs would it take to cover today's ride-share demand? Around 2,000, to hit consistent 2-minute pickups. Ok, that's given today's demand, but Cybercab could create entirely new consumer behavior. It may end up getting priced MUCH cheaper than uber, has a more consistent & private experience, you could do email or watch Netflix on your way to and from work, etc. So what would happen if people started commuting with it, and using it instead of their car on most trips... Around 50,000 cars would do the trick. How does this compare to the number of cars on the road in Austin today? Austin has 800,000 cars on the road, so it's much less, but it makes sense given the Cybercabs would have higher utilization. So what if we extrapolate the math to EVERY major US city? 15,000,000 Cybercabs. How much revenue would this translate to for Tesla? Both in terms of the sale of the car, but also in terms of the fare revenue? Sale of car: $420 billion, but spread over several years to reach this point. Fare revenue: $736 billion PER YEAR....! To put that into perspective, that is more annual revenue than Amazon, Walmart, Google, Microsoft, or NVIDIA. It's 7x Tesla's current entire-company revenue base. This reminds me of how SpaceX became a telecom business more than a rocket launch business, by revenue. Tesla may become a ride-hailing business more than a car business, by revenue too. And all this before we think about what Optimus could do. Insane. 🤯

Today, pressure vessels are extremely safe. But 100 years ago, steam boilers (not nuclear) used to explode once a day. In 1905, one explosion happened right under a Shoe Factory in Massachusetts, collapsing the building, and starting a fire. 58 workers were killed. As a result, Massachusetts wrote the first boiler law, and in 1915 the engineering societies published a uniform set of codes, so a vessel built in one state could be trusted in another. That document became the ASME Boiler and Pressure Vessel Code. Nuclear power arrived four decades later, and nearly those same exact codes could be used. But they had to add a new section: Section III, which governs every reactor pressure vessel in America. Why? Because the old ASME boiler code was just addressing: “Can this steel vessel safely hold hot, high-pressure water or steam?”. For a nuclear plant, engineers had to answer: “Can it keep doing that for decades while being bombarded by neutrons, heated and cooled 1000s of times, exposed to radiation, and serving as part of the barrier that keeps radioactive material contained?” So the N-stamp was born, as a certification that Section III had been satisfied, specially created for nuclear reactor vessels. Reactor vessels have had extremely good safety records, thanks to these early precautions from outside the nuclear industry.

BTW, Aalo only spent around $100M total to get to criticality. That includes our full-scale fuel load, full-scale reactor (no sodium), factory and its machinery, all payroll, regulatory fees, etc. Everything. The people who told you nuclear has to be expensive are wrong.

.@AaloAtomics designed its nuclear reactor to fit on a truck. That constraint shapes the vessel, the machines that build it, and the supply chain behind it. Chapter 2 of my visit to @AaloAtomics' Austin pilot factory. 0:00 A reactor that fits on a truck 0:24 Water jets and vessel size 0:55 Designing around existing machines 1:22 Rolling and welding the vessel 1:49 CNC machining 2:03 Thin steel and shorter lead times 2:45 Inside the pilot factory 3:10 Supply chains and "no unobtainium" 3:59 Designing for manufacturing 4:10 Next: liquid sodium

They may sound scary, but sodium reactors have some of the best passive safety properties in the game. Sodium has more than 400 reactor-years of operational experience, and yet has never had a notable radiation release to the public. Here's what keeps it safe: 1. Low pressure. The sodium pool operates near atmospheric pressure. There is no stored pressure energy to force radioactive material out. 2. High boiling point. Sodium operates around 450°C and boils at about 880°C. That is over 400 degrees of margin. 3. The coolant cannot drain away. The core sits in a double-walled pool, which sits in a reinforced guard vessel. Even with a leak, sodium stays around the core. 4. With metallic fuel (fast spectrum), the reactor slows itself down when it heats up. As fuel temperature rises, the chain reaction slows automatically. Even if the control rods did not insert, the reaction would still stop. 5. Decay heat is small. After shutdown, a 10 MWe core produces so little heat that the vessel can shed it to the surrounding air without pumps. This means you don't need backup power to keep the core cool, in a shut-down scenario. This was the challenge in Fukushima, where they lost backup power. 6. Accidents happen slowly. The sodium pool and the graphite hold a large amount of heat before temperatures rise much. In an accident, temperatures change over hours rather than seconds. This gives time to respond. 7. Sodium traps the most dangerous isotopes. Iodine and cesium bind chemically to sodium. In an accident, they stay in the pool instead of escaping to the air. 8. Cooling continues without power. Hot sodium rises and cool sodium sinks, so flow continues on its own if pumps fail. EBR-II demonstrated this at full power in 1986. 9. Small radioactive inventory. A 10 MWe reactor contains about 1% of the radioactive material in a large conventional plant. 10. The design lineage is proven. Sodium pool reactors have operated for decades. EBR-II ran for 30 years and passed deliberate tests where pumps were shut off and safety systems disabled at full power. The reactor shut itself down without human intervention.

Idaho is so beautiful. Our nuclear forefathers decided this would be nuclear's cradle 50 years ago, and built 52 test reactors here. Now, for the first time in half a century, 3 new ones have turned on. All within a few weeks of each other. Very nice place to build the future.

AGI is here. My friend recently met Canadian politician Doug Ford, but his eyes were closed for the photo. Once upon a time this would have been a missed opportunity. Not anymore.

We need to talk WAY more about the upside of what AI could bring. These are the things that excite me the most, in order of increasing craziness: AI will do all the tasks that we don’t really want to do. UBI will mean you can wake up every day, and choose whether to relax, work, create, or do whatever you want (as long as it's legal). We will use AI to discover new science. This will help us to explore the cosmos and uncover many (all!?) of the mysteries of the universe. Humanity will (re)green the Earth, resurrect extinct species, and terraform deserts, oceans, and other planets. Global warming will not be a concern, given total geo-engineering control. We will stop using most of our land for farming, returning it to nature. We will make healthier, cheaper, and more delicious food using science, and we will stop killing conscious animals. AI will fully reverse-engineer the human brain, ending mental health issues that are out of people’s control, unnecessary sadness and depression, etc. People will use AI to design dreamscapes and lucid experiences, merging neuroscience, simulation, and art. Reality tourism will arise, where humans will visit pre-AI historical simulations (like 1600's Earth), or simulated hypothetical situations. Artists will create feelings as they do songs; multi-sensory emotional compositions you can “step into". Groups of people will link partial consciousness to share thought, empathy, or creativity. Not linguistically, but through a shared vector space. Temporary collective minds will form to create, invent, or explore. People will die only if they want to, and will choose either to live in a physical body, or in virtual worlds (or alternating between both). Bodies could become canvases through synthetic biology. We could see new colors, gain new senses, or live safely in new environments like underwater or on other planets. People will map simulated universes, testing new physics, new constants, new forms of life and other incredible emergent properties. Exploration could become the dominant “profession”. Status will come from originality, kindness, depth, and insight, rather than power or possessions.

"It's important we resolve this issue for data centers, because otherwise Elon's going to do it all in space. I think there's a real advantage to doing a lot of it on Earth, because more energy is more wealth. More energy is more abundance. What @ChaseLochmiller (@CrusoeAI) is doing is a great excuse to deploy a ton of energy all around the country. After a PPA is done with a data center, maybe some of these sites just go onto the grid, and suddenly we have a lot more energy in the country. The public should want a lot more energy. I really hope humanity gets this right, because a lot is at stake. Energy is health. Energy is abundance. And we should want it. Otherwise it'll go into space." - @MattLoszak, CEO & cofounder of Aalo Atomics, at @salesforce Dreamforce 2026.

What happened with Nuclear is the best lesson on the importance of not making the same mistake with AI. Overregulation and vilification will cause us to lose out on huge potential upside.