The need for computing power for artificial intelligence startups has not shown any signs of slowing down. Now, there’s a new GPU from Nvidia that might change the game.
What’s happening:
- Nvidia (NASDAQ: NVDA) has unveiled their new superchip for AI workloads, the HGX H200 GPU
- The new hardware offers both faster memory and doubles the inference speed on large language model training
Why it matters:
- From GPUs to NPUs, artificial intelligence startups are all searching for computing power to build generative AI and power their workloads
- There continues to be a massive shortage of GPU cloud computing access which makes the launch of Nvidia’s new chip even more relevant
Going deeper:
- The H200 GPU will greatly enhance image and speech recognition in AI models
- Cloud computing providers including Amazon, Google and Oracle will be amongst some of the first companies to roll out the H200
By the numbers:
- The H200 GPU has 141 gigabytes of memory
- Nvidia has stated the H200 is capable of a 60-90% performance improvement for AI workloads when compared to their H100 GPU
Market reaction:
- Shares of Nvidia are up +5% since the announcement of the H200 roll out