Extropic is advancing the AI compute model with physics-based computing and challenges Nvidia's dominance in the chip industry.
Extropic, a hardware startup founded by former Alphabet quantum research team members, has secured $14.1 million in seed funding. Extropic is reportedly developing a chip optimized for running LLMs. The technology is described as a "novel full-stack paradigm of physics-based computing" that incorporates concepts from non-equilibrium thermodynamics, hinting at a unique approach to chip design.
Traditional computing models are constrained by physical and complexity limitations. Extropic's breakthrough aims to build a system where computing errors or noise, which has been a significant challenge in quantum computing, are considered an asset rather than a liability. The company has not disclosed detailed technical information but indicated a goal to reduce the electricity required for running AI models and automate certain coding tasks, potentially allowing the computer to self-program for learning representations of the world.
Extropic is advancing the development of their physics-based computing model, which could automate coding tasks and enhance processing of LLMs, potentially leading to major improvements in AI applications.
Contrary to its founders' background in quantum computing, Extropic's product is not a quantum computing chip. The decision is attributed to the scalability challenges of quantum physics-based computers and the company's pursuit of a different path to practical physics-based computing.
This development challenges Nvidia's dominance in AI hardware, especially with its H200 data center processor. Extropic's approach, utilizing physics-based computing, offers a compelling alternative in the competitive AI chip industry.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Article",
"mainEntityOfPage": {
"@type": "WebPage",
"@id": "https://www.themoney.game/stories/former-alphabet-experts-raise-14-1m-for-their-physics-based-ai-hardware"
},
"headline": "Former Alphabet Experts Raise $14.1M For Their Physics-Based AI Hardware",
"image": "",
"author": {
"@type": "Organization",
"name": "The Money Game",
"url": "https://www.themoney.game"
},
"publisher": {
"@type": "Organization",
"name": "The Money Game",
"logo": {
"@type": "ImageObject",
"url": ""
}
},
"datePublished": "2023-12-11"
}
</script>