eu-inc.org → If you care about Europe, please sign this petition to set up an overarching European entity headed up by the EU/Acc movement.
🤼 People
Lars Frølund - Advisor to the NATO Innovation Fund
Lars was on stage at the European Defense Tech Summit and talked about institutional investment into European defense tech. A few hours later we had time to chat about how he got the NATO representatives to agree on funding the NIF. The war in Ukraine was a catalyst for the parties to close in on an agreement, though there are still a few hold-outs (US and France notably).
💼 Portfolio jobs board
This weeks selection of opportunities from the portfolio:
Founding CTO with hardware/EE expertise for sovereign compute company in SF (email me at my name @inflection.xyz )
🚀 Companies
Solcoa - Metal refinement for the future
Status: 600k raised
Source: CAVI
Founders: Hooman Reza Nezhad
Why it’s cool:
If we want living standards to go up, we need to increase global wealth. This is best done by massive breakthroughs in efficiency. Today, metal/alloy working consumes 10% of the world’s energy output. If we use less energy, produce less waste and less waste heat we can achieve a lot more.
LXQ - really fast neural nets
Status: Recently out of stealth
Source: Network
Founders: Georg Meinhardt
Why it’s cool:
Inference is the new training:
CERN Fast Inference Benchmark:
JSC 72 % in 1.3 ns = 0.0000000013 s
Standard AI Benchmarks:
MNIST 99.3 % in 6.1 ns
CIFAR-10 69 % in 6.9 ns
💡 Ideas & Science
Machines of Loving Grace by Dario Amodei
New AI think-piece by the founder of Anthropic making moderate but insanely bullish predictions about the future of humanity (with AI). Dario posits that once sufficiently powerful AI (AGI) is achieved, we could have a “country of geniuses in a datacenter” - meaning the equivalent thinking power of hundreds of thousands if not millions of genius-level, 24/7-working problem solvers, all within the confines of a data center. They will need a physical loop, for manipulating the real world and collecting data/testing hypotheses, but they won’t necessarily need a human in the loop. He lays out five areas where intelligence is a current bottleneck to producing more output: bio & health, neuroscience & mind, economic development and poverty, peace & governance, and work & meaning.
MIT team takes a major step toward fully 3D-printed active electronics
While it’s not the atomic 3D-printing we were promised, it’s at least resettable fuses 3D printed! Not going to be replacing lithography, masks, and CVD anytime soon it seems. But this could be a way out of critical dependencies.
We use about 1/8th as much silicon to make a watt of solar panels as we did 20 years ago.
One of a few really important inflection points for the past decades.
Here’s another one, in case you were wondering: