[ad_1]
The race is on, and the revolution might be televised. Chatter about Groq began on the location previously referred to as Twitter when Matt Shumer, CEO of HypeWriteAI and OthersideAI, waxed poetic about Groq’s AI PU Inference Engine, reporting Mixtral serving at almost 500 tok/s (in different phrases, delivering instantaneous solutions), and shortly after served up an eye-watering demo. Groq CEO Jonathan Ross even visited CNN to indicate off the engine, blowing a reporter’s thoughts.
And it’s all in service of delivering a secure different to Nvidia’s GPU supremacy. Nvidia is betting that with its earnings up 265% from final 12 months, it’s going to hold on to that dominance. Groq says at worst its know-how makes use of a couple of third of the ability of a GPU, however most workloads use as little as a tenth of the ability. And in a world the place it looks as if these LLM workloads won’t ever cease scaling, and power demand will simply continue to grow (whereas enjoyable truth, the earth is teetering on the sting of local weather catastrophe), effectively, Groq is a serious challenger to the GPU-dominated, extremely bottlenecked, resource-starved compute panorama.
In actual fact, Ross claims that by subsequent 12 months, over half of the globe’s inference computing might be operating on their chips. What’s that world going to seem like? Ross may have the solutions and much more at VentureBeat’s Remodel 2024. He’s going to be diving into why AI inference is essential to enterprise know-how, why effectivity is right now’s pressing want — and why Groq is on the forefront of that revolution. He’ll offer a type of mind-blowing dwell demos to focus on Groq’s capabilities, and discuss what’s subsequent on the roadmap.
Register now for unique insights into Groq’s newest developments, and an opportunity to get within the room with the highest AI innovators on the earth and listen to all the new gossip firsthand. VB Remodel 2024 is going on July 9 – 11 in San Franciso. The theme has by no means been extra topical: placing AI to work at scale, with a deal with the real-world questions and challenges, plus the sensible gen AI case research and software tales that matter most, immediately from business leaders.
The countdown is on: Register now!
[ad_2]