OpenAI Chip Beats Nvidia Systems with 1.9x More Work per Watt
- On Tuesday, OpenAI presented benchmark results at the Hot Chips conference showing its custom Jalapeño inference chip outperforms Nvidia's GB300 system in AI work per unit of power and response speed.
- Developed in collaboration with Broadcom, the Jalapeño chip is the first in OpenAI's custom silicon series, designed partly by AI to streamline inference tasks for ChatGPT and future products.
- Testing on the InferenceX benchmark shows Jalapeño delivers 1.5 to 1.9 times more AI work at peak throughput and 1.7 to 3.6 times lower latency, while operating at 700 watts.
- OpenAI plans to deploy Jalapeño in small volumes by late 2026, with significant production in 2027, though the company will continue relying on Nvidia GPUs for compute-heavy model training.
- As OpenAI joins Google, Amazon, and Anthropic in building custom silicon, industry analysts caution that vendor-supplied benchmarks should be verified independently before confirming these performance claims.
34 Articles
34 Articles
(Seoul = Yonhap News) Reporter Jung Joo-ho = OpenAI's self-developed AI chip 'Jalapeno' has outperformed Nvidia's current product line in performance tests...
OpenAI Unveils Jalapeno AI Chip for Faster Inference Than Nvidia GPUs
OpenAI has introduced a custom artificial intelligence chip named Jalapeno that promises quicker response times than competing solutions from companies such as Nvidia. The development marks a significant step in the company's efforts to build its own hardware infrastructure to support the growing demands of advanced AI models. According to a report from Digital Trends, OpenAI claims the Jalapeno processor can deliver faster inference speeds comp…
OpenAI chip beats Nvidia systems with 1.9x more work per watt
OpenAI says its first custom inference chip, Jalapeño, can deliver more AI work per watt while also reducing response times, pointing to a hardware design aimed at handling increasingly demanding model workloads more efficiently. The company tested Jalapeño across three public models: GPT-OSS 120B, DeepSeek R1 670B, and Kimi K2.5 1T. Across the tests, the chip delivered 1.5 to 1.9 times more AI work per watt at peak throughput and 1.7 to 3.6 tim…
OpenAI Claims an AI Chip Breakthrough. Broadcom Collaboration Jalapeño Beats Nvidia's GB300 in Tests.
The new processor, called Jalapeño, outperformed Nvidia's GB300 in the amount of AI work completed per unit of power and the speed at which the system returned responses, according to the company.
Coverage Details
Bias Distribution
- 50% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
