OpenAI claims its Jalapeno chip outperforms Nvidia GB300, promises faster and cheaper AI
Follow UsOpenAI says its new Jalapeno AI chip has outperformed Nvidia’s current systems in key tests.
Jalapeno performed better than the Nvidia GB300.
The company is already working on a second-generation Jalapeno chip.
OpenAI has claimed that its Jalapeno AI chip has outperformed Nvidia’s current systems in key tests. The chip was tested for how much AI work it can handle for the amount of power it uses, as well as how quickly it can respond to requests. According to OpenAI chip chief Richard Ho, Jalapeno performed better than the Nvidia GB300, reports Bloomberg. OpenAI plans to start using its Jalapeno chip to run its AI models later this year. The move may help the company reduce costs and improve the speed of its AI services.
SurveyTypically, some chips are designed to handle more AI workloads, while others focus on delivering faster responses. However, Jalapeno is designed to offer both, Ho said. OpenAI will decide which AI models run on the chip, allowing customers to choose between options that prioritise lower costs or better performance.
Also read: OpenAI faces investigation over Hugging Face hack and alleged lack of oversight
“In the lab, Jalapeno is showing performance both in the high-throughput domain, meaning it will be able to serve a lot of customers more cheaply, as well as the low-latency domain, meaning that for the customers that care about it, the response time will be really, really fast,” Ho was quoted as saying in the report.
For those unaware, Jalapeno was developed by OpenAI with semiconductor company Broadcom. The two companies announced their partnership last year. The chip is designed mainly for AI inference, which is the stage when an already-trained AI model responds to user requests. It is not designed for training AI models, an area where Nvidia’s processors remain strong.
Also read: Sam Altman says he expected AI to change businesses faster, admits he was wrong
One of Jalapeno’s biggest advantages is its power efficiency. OpenAI says the chip can deliver strong performance while using around 700 watts of power. Lower power use can help reduce the cost of operating the large data centres needed to run AI services.
Do note that Jalapeno was not tested against Nvidia’s newest Vera Rubin chips, which has recently started shipping. OpenAI also says it will continue using chips from Nvidia and other providers.
The company is already working on a second-generation Jalapeno chip. OpenAI expects to complete its design in the coming months, according to the report. It has also started planning a third generation as it looks for ways to lower the cost of building and operating its growing AI infrastructure.
“We have so much need for compute, which is why we signed up so many different providers,” Ho said. “That’s going to continue for a while.”
