Sarah Guo

@JeffiniteVC each TRISO particle (the fuel that many SMRs are designed to use) is its own containment system. it is intrinsically much safer to store. there are many paths to nuclear

Open American models! and a really clear, transparent explanation of design and training to boot

@hthieblot that’s quite a claim. would Steve name his company heyclicky

@andrew__reed I can get this sign too

energy is local. stopping datacenter builds, as NY just did (rather than designing win-win requirements and incentives for local communities and builders) is going to lead to some geos having decrepit infrastructure, and some geos benefiting from grid upgrade, investment and economic revitalization. the state needs to “plan” faster. the last time the bureaucracy said they’d study energy it led to a 30-year abject failure to build any (clean) nuclear capacity. a standard nuclear EIS takes 4 years to even get a decision

@pythianism no one asks that

@clayhaight let’s gooo

@traestephens this is a great idea

guys, robots are real I’m so juiced about it

so proud of the amazing team @chaidiscovery! deployment with Eli Lilly, Pfizer and Novartis, and another $400M to build. we are standing in the foothills of the next era of medicine

.@ouraring Tom I would pay for the ability to inflate my own sleep metrics, to myself

my first ever (!) stock pick, in 2011, was $BKNG. this week's @NoPriorsPod with @BookingHoldings CEO Glenn Fogel on marketplaces, travel, AI, and why moats aren't real https://t.co/4d80uZgNG6

Americans want to own homes, build wealth, and have economic opportunity. So some are voting for socialists like Mamdani because he has recognized these anxieties and identified a villain (but not a real solution). What is most likely to help? Socialism consistently makes everyone equally poor. The experiment has been run already! You don’t build prosperity by attacking the entrepreneurs, businesspeople, or investors who create it. We do not want to be Maoist China, the Soviet Union, Venezuela, Cuba, East Germany, North Korea. This is a very bad club of outcomes. The next decade will create more business owners than the last century. AI giv…

cute ad, killer founding team https://t.co/cL6Tj0pxuQ

Very proud to be an American! There is no better system: individual rights, popular sovereignty, limited but capable government, rule of law, pluralism, free speech and markets, earned status, self-correction, invention — and ambition. We will be better for the next 250 by leaning again into who we are, and as an ongoing nation-project, fixing the pieces of that which need work (housing, capacity to build physical things, governance capacity, trust collapse, defense against propaganda). 🇺🇸

On this great birthday, let’s remember the classic American pattern: we get embarrassed, then we get busy. Sputnik went up, and within a year we’d built NASA and DARPA (new institutions, created at warp speed, that gave us the moon landing and the internet). At 250, we shouldn’t need the embarrassment. Crisis got us this far, but ambition must take us from here. We have plenty of it in this country. That ambition (and the country) deserves a government on a shot clock! Let’s be great on purpose now:

Egypt won?

investors buying secondary in a company they can’t get primary access to is quite common — but the investor promoting that online as if they’re a direct shareholder and partner to the company is disingenuous. And will get folks embarrassed by founders “who do not consent” to the new ownership

super impressive example of combining machines-asking-humans in order to efficiently direct use of compute against a goal

nuclear powered GPUs!!

Many parts of venture don’t make any sense to founders, and hanging out with the great but always surprising @stanine, being forced to explain how VC works, reminded me of that

the destruction of American education over the past decade is an incredible self-own competence is objective. a child can either do the math or they can’t. but in the u.s., a lot of people have reasons not to say that! parents don’t want to hear their kid is struggling. teachers don’t want scores used to manage them. districts don’t want embarrassment. progressives worry accountability will create inequity. conservatives don’t want federal authorities. so we end up with process, weak standards, and excuses to explain away bad outcomes. people object that it’s phones, covid, demographic change. ok! but we fail globally when others have pho…

has anyone hosted a debate on solar+batteries vs nuclear for datacenter power generation? (both would be good, but curious) before someone @ me, yes I understand that there are cloudy days sometimes and they have different generation profiles and solar might need overbuild/diesel backup etc. that’s the point of the debate

Really fun to hang again with my friend 🃏 @polynoamial (OpenAI research scientist, our first guest ever on @NoPriorsPod in early 2023) to talk about the implications of large test-time compute, and what happens when models are given $10M budgets to spend on a single task. Topics: 01:23 – Why Benchmarks Are Broken 04:19 – Compute Budgets and Projections 06:48 – How Long Should Models Think? 08:01 – Benchmarkmaxxing 09:48 – Noam's Evals 12:40 – Safety (When Model Capability Scales With Spend) 16:09 – Implications For the Model Release Cycle 18:34 – Latent Model Capability 22:27 – Limits on Recursive Self-Improvement 28:38 – Large-Scale Multi-…

Cool research work on scaling inference compute

I start by assuming the best in people, but I also follow the basic rule of thumb — when people show you who they are, believe them the first time.

didn’t think this would happen for a while, but now I consistently downgrade model for chat queries I know are not that hard, should be mostly retrieval/ai search api bc I want faster speed cc: @KeenableAI

@iyanmoonyang @ryancjulian even worse in robotics than llm. eval depends on hardware, objects, environment (scene, lighting, calibration), policies / recovery behavior… pairwise preference feels tough to do

ICYMI, full @NoPriorsPod discussion at @Microsoft Build with @satyanadella 01:48 – Reflections from Microsoft Build, ecosystem 03:12 – MAI Training, Model Strategy 05:48 – Complexity of Real-World Deployment of AI 07:33 – Augmenting Human Capital 09:37 – Harnesses for Enterprise 11:49 – Value Capture By Developers 15:09 – Can Everybody Operate at the Frontier? 15:51 – The Definition of IP in 2026 17:38 – Vendor vs. Enterprise Agents 21:48 – Predictions on Model Pricing 24:02 – Durability of SaaS, SaaSpocalypse 25:58 – What Satya’s Coding Himself 28:18 – Future of Engineering 30:54 – How Microsoft Can Be More Ambitious? 34:36 – Data Centers a…

anti-datacenter is just de-growth by another name instead of refusing to enable humans with more capabilities, we can figure out how to make more energy, mitigate electricity cost, and get to zero climate impact orbital datacenters becoming a legitimate part of the discussion right now is existence proof. we would be pulling power out of the sky. more short term, the needs of datacenters are a catalyst for building industrial capacity (including power generation) we should have built in America long ago. technology (and regulatory reform, redistribution) is the answer to abundance degrowth is an inhumane fantasy of folks who do not unders…