OpenAI's Jalapeño Chip Surpasses Nvidia's Blackwell Ahead of Earnings Report
OpenAI has unveiled its Jalapeño chip, claiming it achieves up to 1.9 times more inference throughput than Nvidia's Blackwell chip. This announcement arrives just before Nvidia is set to report its fiscal second-quarter results, which analysts expect to show a doubling of revenue year over year. The Jalapeño chip, co-developed with Broadcom, is not expected to enter volume production until 2027.
According to OpenAI's benchmarks, the Jalapeño chip delivers between 1.5 and 1.9 times more AI work at peak throughput compared to Nvidia's Blackwell systems. It also boasts significantly lower latency, with improvements ranging from 1.7 to 3.6 times in end-to-end latency and 2.1 to 4.1 times faster ultra-low-latency interactive inference across various AI models. Each rack of the Jalapeño chip contains 128 accelerators, providing 1.7 exaFLOPS of compute power and nearly 2 petabytes per second of memory bandwidth. Richard Ho, VP of Hardware at OpenAI, stated that the chip achieves high throughput and low latency simultaneously, marking a significant industry milestone.
For Your Trading Desk
Sponsor Spike Success.
However, it is important to note that the Jalapeño chip is designed solely for inference tasks, and Nvidia continues to maintain a stronghold in training capabilities. The comparison specifically targets Nvidia's GB200 and GB300 racks, which were launched in 2024 and 2025, respectively, and does not include the upcoming Vera Rubin platform. OpenAI has not disclosed system-level power metrics, meaning the benchmarks focus on throughput rather than performance per watt.
Nvidia's earnings report is anticipated to show revenue between $91 billion and $92 billion, with earnings per share around $2.09. Despite consistently exceeding consensus estimates in recent quarters, Nvidia's stock has experienced declines following four of its last five earnings reports. Currently, Nvidia shares are trading at $213.05, reflecting a 3.04% drop for the week but a 14.37% increase year to date.
The competitive landscape for AI chips is intensifying, with major players like Google and Amazon also developing their own chips. Analysts have highlighted potential risks for Nvidia, including a projected 4% decrease in GPU market share over the next few years. OpenAI's announcement adds another layer of complexity to Nvidia's market position, especially as its largest customers increasingly become competitors.
As Nvidia prepares to release its earnings, investors will be watching for any shifts in data-center guidance and how the company addresses the growing competition in custom silicon during its earnings call. The implications of OpenAI's Jalapeño chip could have lasting effects on Nvidia's market strategy and investor sentiment in the tech sector.




