Charles Yang
@charlesxjyang
Building @industrialstrat, theology @GordonConwell, Founder @RVA_YIMBY + @RVA_EBikes || prev @ENERGY, @SambaNovaAI, @ARPAE
Huge congrats to @DeptofDefense for aligning the offices, funding lines, and authorities to make this deal happen. Now the hard part: 10+ years of implementation and project management by DOD, which will determine whether this project fails or suceeds My one piece of advice…
So proud 🇺🇸 MP Materials Announces a Transformational Public-Private Partnership with the U.S. Department of Defense – Multibillion-dollar DoD commitment – New “10X” magnet facility to be constructed – 10-year price floor + offtake agreements – DoD positioned to become MP’s…
We’re launching Right Of Way, a podcast about energy policy, energy politics, and above all the upcoming permitting reform fight. Hosted by me and @ppavnr, presented by @IFP and @JoinFAI. First episode dropping next week — stay tuned.
If @ppavnr and I hosted a limited series podcast all about the upcoming permitting / highway bill fight, would people listen?
The great flaw in our system of Government is not the temptation it offers the strong man, but the latitude it allows the weak man to do less than is necessary.
INTRODUCING the Rickover Corpus: >2500 pages of never before seen speeches, memos, and congressional testimony from Admiral Hyman Rickover On topics ranging from education, defense procurement, nuclear engineering, politics, and more
New podcast episode out with Ken Ono, professor of math at @UVA, who's been working with @EpochAIResearch on creating math benchmarks for AI. Substack link attached
All I know is the king of hydro @charlesxjyang is partying hard tn
Google plans to buy up to 3 GW of hydro energy from Brookfield... Agreement includes $3 billion in contracts for power from two dams in Pennsylvania. Ability to relicense, overhaul, upgrade assets to extend useful life. World's largest corporate deal for hydroelectricity.
Kierkegaard wrote of faith, but he could’ve been describing a founder:
Another paper showing that in physical domains, general-purpose models often learn shortcuts—not underlying structure—unlike inductively biased models like ODEnets. Which is why I'm excited to keep digging into how we build better scientific models at @InSilicoPod!
Can an AI model predict perfectly and still have a terrible world model? What would that even mean? Our new ICML paper formalizes these questions One result tells the story: A transformer trained on 10M solar systems nails planetary orbits. But it botches gravitational laws 🧵
Biden may have been a great oil trader but Trump's DoD just booked $200M in equity gains from this MP Materials deal. Direct investments like this can help the government offset the fiscal cost of price floors,
$MP MP Materials Shares Jump on $400m Defense Department Deal