Our 2022 annual letter is out, including our views on the year past and how it shapes how we see the future... chamath.substack.com/p/2022-…

We have terminated the employment of two of our employees due to employee-specific circumstances. We have no further comments at this time.

Congrats @JonathanRoss321 @sundeep - we’ve been working on this for some time and it’s already started producing results. Looking forward to seeing even more accelerated #GroqSpeed for #GroqCloud

I will be hosting a Q+A live on Spaces today at 2:30pm PST. If you are interested in joining in and asking questions, you can set a reminder below: x.com/i/spaces/1LyGBnDzPZLGN

Monster LPUs being racked up to serve the more than 3,000 pilot customers! 💪 #GroqSpeed

Saildrone’s celebrating a major #milestone! Our fleet of #USVs surpassed an incredible 1M nm over 32,000 days.💪 “Evolving the Saildrone design, capabilities, and team has been an incredible journey,” said Saildrone founder and CEO Richard Jenkins. saildrone.com/news/saildrone…

SD 1045 set the #GuinnessWorldRecord for highest wind speed recorded by a #USV during its 24 hours in Hurricane Sam. Measuring a maximum of 109.83 knots (126.4 mph), the uncrewed surface vehicle will be in the @GWR 2024 issue. 🏆💪#hurricanedata @NOAA saildrone.com/news/guinness-…

Ever wonder what it is like to be in the center of a hurricane? Video from @saildrone (uncrewed ocean robot) captured the inside of Hurricane Idalia last month in the Gulf. More: weathernationtv.com/news/vid…

In the arena with super amazing @chamath 😇 big supporter of work @HarvestingFN is doing with small scale farmers in India 🚀🚀🚀 Leave No Farmer Behind #HFN

Leave No Farmer Behind #HFN bringing food security to the world one day at a time!

With super amazing @superjz @socialcapital 🚀🚀 board member and true partner at @HarvestingFN Leave No Farmer Behind #HFN

If you, or an adult you care about, may have ADHD, there is an FDA authorized technology (ie non drug option) for treating ADHD that is now in a product you can get without a prescription. And unlike stimulants, there are very low side effects and the effects appear to last when you’re not using the product. Please read below if curious why it’s grown to a Top 20 Medical app since launch in June. @EndeavorOTC helps you with cognitive strengthening through a game using an algorithm that acts as a personal trainer, tailoring attention challenges to individual users. A multi-site, 200+ participant trial yielded very strong results in treating ADHD - 83% experienced a clinically meaningful change in attention function as measured by the TOVA ACS. 73% of participants also reported quality of life improvements as measured on a validated quality of life scale. See below for the trial summary slide. Separately, no serious side effects have been reported. For full use and safety information see Instructions for Use: endeavorotc.com/IFU. You can start a free trial or subscribe here if interested: apps.apple.com/us/app/endeav…. Disclaimer: Akili, the maker of this product, is a company we invested in to develop digital therapeutics to target these and other issues in the brain.

Llama2 70B - blazing fast, low wattage, low cost @GroqInc LPU (language processing unit) running on Groq racks put together in days. High T/s, low T/watt, low $/T specifically for inference.

“Both companies will field very interesting offerings available in early 2024...upstarts like @GroqInc have shown advantages on certain models and workflows for LLM inference, today,” Moorhead added. #Groq outperforms NVDA in most #AI finance.yahoo.com/news/nvidi… via @YahooFinance

The AI compute market will all boil down to T/s (tokens per second). So if you’re trying to figure out whether to buy $NVDA or not, consider the below… When Web2.0 exploded in the late 2000s, AWS was principally responsible for arming the rebels - they offered startups compute (EC2) and storage (S3) in bite sized pieces we could pay for with a credit card. As a result a thousand flowers bloomed. This will happen for AI compute clusters. And as it does a few things should happen: 1) startups won’t care what chips are used - a100, h100, Groq etc. 2) CUDA lockin will be broken 3) foundational model diversity will emerge As all of this gets figured out, it’s generally good for incumbent revenues but generally bad for incumbent enterprise values (see Cisco c2000s). It’s also very good for emerging disrupters like @GroqInc (disclaimer: I helped get this company off the ground 8 years ago and am one of their largest external shareholders). I expect startups like @GroqInc to push the market aggressively towards super cheap T/s: they should show that they can drive massive throughout at the lowest relative cost. The cloud providers will then do the rest…

TL;DR: If you’re too lazy to read this, then just look at the graph! We need to be in the upper left quadrant for US energy independence, national security and mass electrification. The problem with mass electrification in the West today is it relies on complicated materials like Nickel and Cobalt. These are not broadly available; a lot of it comes from high conflict areas, some of it is mined using child labor, the climate implications can be terrible, and China has stockpiled and will rate limit its use. And these are just a few of the problems… All of this means that batteries are expensive and/or don’t deliver the energy density we want. Part of the reason we are here is that experimentation in battery metals is poor and most of the market is based on technology that is decades old. At the beginning of the pandemic in 2020, I started a conversation over @X with @VivasVK7. That back and forth led to us starting a company: Can we build something that can better predict how different chemical elements may work together to make better battery materials? By building such a system, experiments could run much faster than the counterfactual, and learnings would grow quickly. In current state-of-the-art battery materials, it takes 12-18mo to evaluate a hypothesis about a new chemical composition for metrics like energy density. The company we started, @Mitra_Chem, has shrunk the testing cycle by 90%, which should become even faster and smarter over time. It has allowed us to land on a commercial formulation for an improved version of LFP - called LMFP - that will be scaled into OEM supply chains. IOW, slowly creeping towards the magic quadrant. We have also found a new experimental formulation, LMX, that can deliver the energy density of NMC and NCA at the pack level but at the cost and reliability of LFP. There is still a lot of work to do to make LMX work, but we are hopeful. Aka, sweet foreplay with the magic quadrant is at hand. Anyways, we just raised our first round of external capital, led by @GM. Hopefully they become customers tomorrow as well as investors today. If you are a machine learning/AI engineer, a materials scientist or a chemist, please consider working with us: mitrachem.com/join-us cnbc.com/2023/08/16/gm-inves… Onwards.

Mast friends and family, help us secure a panel at next year’s @SXSW Cast your vote to hear from our CEO @GrantCanary and other climate leaders at SXSW on how to restore resilient forests. 👇Click to vote hubs.li/Q01_clPj0 Thanks for your support.

How the Inflation Reduction Act Has Reshaped the U.S.—and The World time.com/6304143/inflation-r…

General Motors is investing in a startup working on better and cheaper EV batteries cnbc.com/2023/08/16/gm-inves…

GM invests big bucks in battery startup in a bet on new chemistry electrek.co/2023/08/16/gm-in… by @scooterdoll

GM invests big bucks in battery startup in a bet on new chemistry — @ElectrekCo electrek.co/2023/08/16/gm-in…