Breaking News
Windows 10 is DEAD in 2025? -- Here's How I Run It SAFELY Forever (No Updates)
GENIUS ACT TRIGGERED: The Biggest BANK RUN in History is COMING – Prepare NOW
European Billionaires Funneled $2 Billion into NGO Network to Fund Anti-Trump Protest Machine
Japan Confirms Over 600,000 Citizens Killed by COVID mRNA 'Vaccines'
Top Tech News
HUGE 32kWh LiFePO4 DIY Battery w/ 628Ah Cells! 90 Minute Build
What Has Bitcoin Become 17 Years After Satoshi Nakamoto Published The Whitepaper?
Japan just injected artificial blood into a human. No blood type needed. No refrigeration.
The 6 Best LLM Tools To Run Models Locally
Testing My First Sodium-Ion Solar Battery
A man once paralyzed from the waist down now stands on his own, not with machines or wires,...
Review: Thumb-sized thermal camera turns your phone into a smart tool
Army To Bring Nuclear Microreactors To Its Bases By 2028
Nissan Says It's On Track For Solid-State Batteries That Double EV Range By 2028
Carbon based computers that run on iron
Tesla Dojo, Supercomputers and AI
There does not seem to be a limit for neural nets to utilize more resources to get better and faster results.
Tesla is motivated to develop bigger, faster computers that are precisely suited to their needs.
The Google TPU architecture has not evolved as much over the last 5 years. The Google TPU chip is designed for the problems that Google runs. They are not optimized for training AI.
Tesla has rethought the problem of AI training and designed the Dojo AI supercomputer to optimally solve their problems.
If Tesla commercializes the AI supercomputer that will help to get to lower costs and greater power with more economies of scale.
One of the reasons that TSMC overtook Intel was that TSMC was making most of the ARM chips for cellphones. TSMC having more volume let them learn faster and drive down costs and accelerate technology.
99% of what neural network nodes do are 8 by 8 matrix multiply and 1% that is more like a general computer. Tesla created a superscalar GPU to optimize for this compute load.