Today, China dominates rare-earth mining and processing. But that wasn’t always the case. Solcoa’s @hoomanrenezhad explains how the West lost its rare-earth mojo: "So contrary to popular belief, we do mine in the West. We have an enormous rare earth mine in Australia. There’s one in California." “The actual raw feedstock is not rare." “The bottleneck is what you do after it. It’s the processing. That is something that we’re completely unable to do in the West." “We actually invented these processes in America." "The technologies used to make rare earths in China today are American technologies from the Manhattan Project.” “We built this technology and these systems, and then we shipped it to China. China dominated. They matured it. They took over.” Now his company is trying to bring that capability back to the U.S. with a process built for today’s economics. “If we were just copying that same technology stack again, it was impossible. The economics don’t close.”

In a warehouse in Alameda, I watched @HoomanReNezhad hold a crowbar up to a magnet that used to run an MRI machine. The magnet won. It tore the bar out of his hands and pinned it to itself with a loud crack. That magnet is about to be melted down to recover its neodymium, the metal at the heart of every EV motor, drone, wind turbine, and precision weapon America builds. You can't make one of these magnets without it. And almost no one outside China can turn rare earths into the pure metal the magnets require. Solcoa Industries can. Hooman and his co-founder Artem Iurkovskyi have reinvented the chemistry that turns rare earth and scrap into magnet-grade metal. They do it at lower cost than the conventional route, without the greenhouse gases. The US hasn't commercialized a new way to make a primary metal since the 1940s. Solcoa is doing it today. A year ago they made this metal by hand, a few grams at a time. Today the line runs at 10 tonnes a year and their first commercial plant is built for 500. America wins by inventing. @Felicis is proud to invest in Solcoa's Seed.

Demand for magnet-grade rare earth metal could grow four to seven times by 2040, and today most of it is made in China. Solcoa Industries has reinvented the chemistry to make it in America, and is scaling it in a warehouse in Alameda. Why we backed @hoomanrenezhad and Artem Iurkovskyi, from @DetweilerJames: felicis.link/8pjQ7je

Today we're announcing $75M to build Solcoa One, our 500-tonne rare earth metallization plant in Nevada. At Solcoa, we believe rare earths are the foundation of America's next great industrial era. An era we intend to win on the merits, with superior American technology.

Today's self-play breakthrough of S1 learning soccer from scratch with zero demos is the AlphaGo moment for physical AI. We at @Felicis have been true believers in @SkildAI since Series A. Huge congrats @deepakpathak, Abhinav Gupta & team 🤖 ⚽

Today's self-play soccer result (140+ sim years, zero demos) is the Alpha Go moment for physical AI. We at @Felicis have been true believers since the start. Congrats @deepakpathak & Abhinav Gupta, and the entire @SkildAI team 🤖 ⚽ cc: @asenkut

We trained a robot to play football. How? Self-play for 140 years in a virtual World Cup. Meet the #Messinator:

Oh hi @mercor @datologyai 🧡

Today we're launching DIFFUSE: a platform for agencies, brands and studios to search for, connect with and hire AI-native talent. There's a lot of worry around AI displacing creatives but the reality we keep seeing is different: an entirely new creative workforce is emerging, and demand for it is accelerating. DIFFUSE connects that talent with this new landscape of opportunities. Register and learn more at the link below.

🔥 The latest from Felicis: - @RadiantNuclear landed a U.S. Army agreement worth up to $750M to deploy 15 portable nuclear microreactors. - @SkildAI taught a robot a 10-minute task from a single video with no fine-tuning. - Heaviside Industries Industries raised a $60M Series B and partnered with Nammo. - @smackaistack raised a $61M Series B to bring tactical AI to the edge. - Magentic is putting AI agents inside the world’s largest manufacturers and announced an $18M Series A. - @AtalantaTech and Artemis Security are pushing cyber defense to machine speed. - @paraformtalent has already helped Palantir hire 50 engineers. - @RicursiveAI co-founders @annadgoldie and @Azaliamirh and @Mercor co-founder @BrendanFoody were named to the TIME 100 AI list. And we’re thrilled to welcome @_tonyshi to @Felicis as our newest investment partner, backing ambitious student and researcher-founders from day one. Stay up to date on all things Felicis by subscribing to our newsletter: felicisnewsletter.substack.c…

OCT 6, San Francisco @pxd and I are bringing the design chiefs of Anthropic and OpenAI @joellewenstein and @iansilber on to one stage for a 4-way convo about AI, agents, and the future of design There will be great coffee, great banter, and great company. Hosted by @felicis

today, hundreds of thousands of people interact with an archil disk every week without realizing it. because archil is the easiest way to build stateful agents and data platforms. this week, for the first time, we exceeded 1 mount per second for the entire week. at this rate, in four weeks, we will be creating more disks each week than supabase creates databases. each archil disk can run databases, git checkouts, npm, and s3-like storage -- supporting tens of thousands of operations per second. leggo

The physical world is getting a digital workforce. Magentic is building the digital workforce for manufacturing, supply chain, and finance. Its AI workers, called Mages, diagnose problems, gather evidence, plan the next step, and act across fragmented enterprise systems, while people stay in command. Companies including Coca-Cola Europacific Partners and Siemens are already working with Magentic. We at @Felicis are proud to lead Magentic’s Series A. Congratulations to Robin Van Aeken, Odhran O’Donoghue, and the entire Magentic team. The physical world is ready for agents. cc: @speechu @FHaskaraman felicis.link/89PPsGd

I spent years advising companies on supply chain and saw how much critical work depended on fragmented data and manual effort. With Magentic, the physical world gets a digital workforce. We at @Felicis are proud to lead the Series A. felicis.link/3Qvmvj1

1/ Robots have a data problem. Every major data wave has created infrastructure companies designed to handle the specific use cases of that era. Relational. Web. Real-time. Physical-world data is next. Who defines its stack is still up for grabs. Essay for @felicis: felicis.link/Drumsjx

Relational data. Web-scale data. Real-time analytics. Every era of data has produced, by necessity, new platforms and companies to service it. Robots are next. Robots produce terabytes across sensor streams that must stay synchronized to the millisecond. The physical-world data layer is still up for grabs. @tecoker's new essay on the physical world's missing data stack: felicis.link/q07LKDY