Michael Staton @mpstaton

Stills from @WHOSTP Director Michael Kratsios touring Aalo's CTR Facility, marking the 1-year anniversary of U.S. nuclear EOs.

keyboard_arrow_left Previous keyboard_arrow_right Next

U.S. Department of Energy @ENERGY

3 weeks ago

x.com/i/article/2058…

21 92 516 46K 74

Episode 3 of "Critical Thinking" is live: Inside the Aalo-1 Core. @yasir_fission sits down with Aalo engineers (and resident PhDs) Firas Abdullatif and David Jaluvka for a dive into reactor physics, the Aalo-1 core, and the designs choices that went into it. Firas and David walk through Aalo's combination of UO2 fuel, liquid sodium coolant, and a graphite moderator, and the tradeoffs that came with each. You'll also learn about reactor physics from the ground up: how fission actually works, fast vs. thermal spectrum reactors, why HALEU supply chain matters, and how neutronics, thermal hydraulics, shielding, and manufacturing all have to come together to make a reactor real. If you caught Episode 2 (sodium coolant), this picks up where that left off. Link: youtu.be/7DaZw_fzD2E

@claude_code I know you have that "how are we doing" and "share conversation with Anthropic" feature. But I'm pretty sure if you just have a trigger for a user saying "I love you" to Claude Code, you should be asking for the agent traces.

@claude_code is my boo.

In the 1980s, the world added ~230 GW of nuclear capacity. (in 10 years). A new reactor connected to the grid every 17 days on average. Insane.

keyboard_arrow_left Previous keyboard_arrow_right Next

@MattLoszak amazing work, on schedule and ahead of expectations at the same time.

Aalo has received DSA approval from DOE for our 10 MW reactor, zero-power criticality! This is the equivalent of an NRC license in DOE space. We've been working towards this milestone at Aalo for 2.5 years. This was a very ambitious project under the DOE reactor pilot program for a bunch of reasons: ➡️ 10 MW, commercial-scale system. This is a system with the full-scale amount of fuel, moderator, controls, etc, necessary for powering AI data centers. Proving out this supply chain is a big deal. In the coming weeks we'll be taking it to zero power criticality, and then going to full-power operation in the months that follow. We'll be getting the factory set up in parallel to mass-produce it. ➡️ Built everything ourselves, from scratch. We started with a green field, and within 4 months finished the building, facilities, policies & procedures, reactor, shielding, instrumentation, control, etc. People thought this would be impossible to do on this timeline for a reactor of this size, but we've done it. Hyperscalers care about speed, and we're showing that moving fast is possible at commercial scale. ➡️ Achieving criticality before July 4th. We set out to ensure we met the president's order, and we are on pace to do so. We only started hiring at Aalo 2.5 yrs ago. It was questionable how realistic it was to take on all of the above while being one of the younger nuclear companies, but we are on track to pull it off. Insanely proud of the team for all the hard work, nights and weekends that went into this. Full blog post will be linked in comments.

We need a @YouTube alternative that has a hard no AI slop policy before we can’t differentiate between AI and human content. Once we can’t, we are at the endgame.

@MattLoszak Yes. Much needed, the largest constraint no one can see ;)

Aalo is making a fundamentally new nuclear product: "Affordable, small" nuclear reactors. First, the US made "expensive, small" reactors (1950-1955). One-off, test reactors to learn the tech. Next, "affordable, large" reactors (1955-1990). Demand came from one-off, large utilities. The best way to make nuclear cheap was to make it big and bespoke. Then, nuclear stalled. We forgot how to build nuclear (1990-today). Today, the US makes "expensive, large" reactors. In the last 20 years, we've built just two. But there is one quadrant no country in the world has never achieved: "Affordable, small" nuclear reactors. What's stopped us? There was never the right type of demand. Now, AI data centers require a 100 GW build-out, of a consistent product. And small nuclear is ideal. The key difference this time: A higher willingness to pay. 15 c / kWh, at scale, in return for: speed. This opens the door for mass-production of smaller nuclear reactors to get off the ground, and come down the cost curve. As long as they can be deployed quickly. Aalo is proving this: We will have gone from founding to fission in under 3 years. Once < 10 c / kWh is achieved, this will be a world-changing technology. It will unlock entirely new markets that were not previously accessible: Industrial process heat, desalination, small utilities, EV-charging, ammonia production, and more. For all of these markets, >15 c / kWh would be too much. But below 10 c / kWh, and these are all within reach. A 1 GW-scale reactor is too big, and too slow. A 1 MW-scale reactor is too small, and too expensive. But a 10 MW reactor, mass produced, deployed in 50 MW Pods, could be the "sweet spot". That is Aalo's product. The Aalo Pod. Sodium coolant unlocks 2-5x more power than water or gas cooled reactors of the same size. UO2 fuel unlocks a 40 GW existing supply chain, avoiding TRISO or HALEU bottlenecks. A sub-12-foot vessel diameter unlocks mass transportability. Every decision Aalo makes is centered around manufacturability at extreme scale, to achieve low costs and rapid deployment. And maybe, just maybe, this product will achieve a lower cost than even the record-holding "affordable, large" reactors from the 70s, if sufficient manufacturing scale can be reached. We're going to find out just how low cost nuclear can get.

keyboard_arrow_left Previous keyboard_arrow_right Next

The US is incapable of a sustained campaign in Iran. The geography and the million man IRGC, plus their ahead-of-the-curve commitment to Drones means the US is incapable. You know who could do it? Ukraine. Just wire them 100B and they will handle it.

This is what democracy looks like. No Kings Protest and March in Austin TX, Oct 2025. A protest for the power of the people, support of the constitution, and NO KINGS since 1776 in America. #NoKings #Protest

That time @claudeai had an outage affecting me for 15 minutes and realized I couldn't do anything without it.

@jefielding Nice. What's that Decemeberists' song from way back when? "And the wanting comes in waves....."

@StephNass I want!

@hthieblot A politician?

I talk to thousands of LPs every year. Here's what I'm seeing right now when it comes to how they're approaching VC. 1) Late-stage co-investments have never been hotter. The top 5-7 names have effectively *unlimited* demand. And that supply/demand imbalance has created three real issues: First, the question of *true* access. Many of the SPVs floating around for these companies are not sanctioned by these companies. Buyer beware. These companies are tight on their cap tables, and access can be gated beyond just capacity. The fee creep is getting egregious. Someone shared with me recently that for a top AI lab, a group was charging 15% upfront, 20% carry, and 30% over a 2X. At those economics, you can take what would be a generational company but a bad investment. Fees are the silent killer of returns in late-stage co-invest, and this risk is now being focused on the logo, but ignoring the fee effects. Additionally, for those who aren't getting access to the top companies must go a tier (or two) below. Very dangerous in a high-intensity / valuation market for AI. As I've mentioned, there will be a lot of expensive mistakes made as we figure out what winners will look like in the future. 2) Capital is flowing to large, established brands. This is the clearest trend right now. LPs are concentrating on the top multi-stage and Series A firms. The logic is straightforward: these are easier to underwrite, and given how extreme the power law has become, investors want exposure to the trophy assets soon after investing. If you're a top 10-20 brand, you're oversubscribed, sometimes in multiples. If you're not, the fundraising environment is a different world. 3) Emerging and emerged managers are in the toughest spot, with a notable exception. Spinouts from top firms and operators with strong brands are moving fast and generally very oversubscribed. For everyone else, the bar has never been higher. I think this is actually where the opportunity is for LPs If they have the time/expertise to do so. The issue isn't that LPs don't know this. Most do. The problem is twofold: the time and difficulty of sourcing and diligencing emerging managers is real, and the hangover from 2019-2021 is still very present. A lot of people tried VC during that era who probably shouldn't have, and the failure rate on those Fund 1s has made the entire segment harder to navigate. Matching supply and demand here has become nearly impossible. Supply (Managers) far exceeds demand due to the issues noted.

@CitiesSkylines so why does this game still not work on Linux? Insane.

@AaloAtomics That's funny slop.

🎄⚛️