Back to News
ProductAI Understanding briefing

OpenAI’s Jalapeño ASIC built for internal use, but rollout could expand

OpenAI’s hardware chief says its new Jalapeño inference chip will primarily serve the company’s own workloads, yet leaves open the possibility of broader availability.

4 min readRead the original reporting
Source-provided image accompanying OpenAI’s Jalapeño ASIC built for internal use, but rollout could expand
Attributed reportingSource recorded
Publisher
tomshardware.com
Source link
tomshardware.comhttps://www.tomshardware.com/tech-industry/artificial-intelligence/openais-custom-jalapeno-ai-inference-asic-is-for-openais-internal-use-but-company-leaves-the-door-open-to-broader-rollout-firm-says-it-will-have-its-hands-full-with-jalapeno-for-a-good-long-time
Source type
Reporting by a news outlet — not a first-party document.

What we could not confirm independently: This claim is attributed to the named outlet. We did not verify it against a first-party document. (tomshardware.com)

ContextUnderstand this in 60 seconds

Start here

Key terms

Benchmark
A standardized test or dataset used to measure and compare model performance.
Inference
The runtime phase where a trained model generates predictions or outputs.
Compute
The processing resources required to train and run models, often measured in FLOPS or GPU hours.
Test yourselfAI Models Explained Quiz

What happened

OpenAI unveiled its custom AI ASIC, named Jalapeño, at Hot Chips 2026. According to Richard Ho, OpenAI’s Head of Hardware, the chip is designed to meet the company’s rapidly growing demand. Ho told Tom’s Hardware that OpenAI expects to keep the chip’s capacity focused on internal workloads for the foreseeable future, but he added that the design could theoretically be offered to external parties. The company presented performance benchmarks that compare Jalapeño against Nvidia’s Blackwell GPUs, indicating competitive latency and throughput for inference workloads. The roadmap outlined a multi‑generation plan, suggesting ongoing development beyond the initial release.

At the Hot Chips 2026 conference, OpenAI presented its Jalapeño ASIC, describing it as an ‑focused accelerator built to satisfy the company’s internal needs. The chip is positioned as a response to the growing demand for efficient processing of large language models and other AI workloads.

Richard Ho, OpenAI’s Head of Hardware, emphasized that the company’s priority is to ensure its own requirements are met before considering external customers. He stated, “We have such a strong demand for compute within the company… it’s going to take us a good long time to even fill our own demand.” However, Ho also noted that the design could be used by others, leaving the door open for a broader rollout.

OpenAI provided comparisons showing Jalapeño’s latency and throughput matching or exceeding Nvidia’s Blackwell GPUs on selected tasks. The company outlined a multi‑generation roadmap, indicating plans for future iterations with improved performance and efficiency.

The article includes a transcript of Ho’s interview, but no detailed specifications, pricing, or availability dates were disclosed. OpenAI’s broader hardware strategy remains opaque, with the Jalapeño ASIC representing the first public glimpse of its custom silicon efforts.

Source details: tomshardware.com ↗

Why it matters

OpenAI’s move into custom silicon signals a shift in the AI hardware landscape, where leading model providers are seeking tighter control over the stack that powers their services. By developing an in‑house ASIC, OpenAI can potentially reduce reliance on external GPU suppliers, improve cost efficiency, and tailor performance to its specific model architectures. If the chip is later made available to other firms, it could introduce a new competitor to Nvidia’s dominance in AI , affecting pricing and market dynamics. The announced performance claims, if validated, also provide a for future hardware comparisons and may influence how other companies design their own AI accelerators.

The development underscores a trend where AI companies are internalizing hardware to reduce dependency on third‑party GPU providers, potentially reshaping the AI supply chain.

If OpenAI eventually offers Jalapeño to external customers, it could provide a cost‑effective alternative to existing GPUs, influencing market pricing and prompting competitors to accelerate their own ASIC programs.

The performance claims, if independently verified, could set new benchmarks for efficiency, impacting how future AI models are optimized for hardware.

OpenAI’s roadmap suggests a commitment to ongoing hardware innovation, which may affect future model releases and the overall cost structure of AI services.

Interactive Mechanism

Interactive Mechanism: How It Actually Works

Explore the underlying technology behind this development interactively.

Model Parameter Size:8B Parameters
VRAM Required5.5 GBGPU memory footprint
Target HardwareMacBook / Single GPUDeployment tier
Privacy100% Air-GappedLocal device capability
Core takeaway: Small, quantized models (3B–8B) now run directly inside smartphones and laptops with complete data privacy, while mammoth 400B+ models remain the domain of datacenter clusters.
Interactive Concept Check+10 Points
AI Models Explained Quiz

In AI, what are a model's "parameters"?

What to watch next

Key points to monitor include whether OpenAI releases detailed performance metrics beyond the initial , any formal announcement of external availability or pricing, and how the chip’s architecture compares to Nvidia’s Blackwell in real‑world workloads. Additionally, the response from GPU vendors and potential partnerships with hardware manufacturers will shape the competitive landscape. Finally, any updates on the multi‑generation roadmap could indicate OpenAI’s long‑term commitment to proprietary hardware.

Whether OpenAI publishes detailed performance data, including power consumption and real‑world workload results, to substantiate its claims.

Any formal statement regarding external availability, pricing models, or partnership opportunities for the Jalapeño ASIC.

Reactions from Nvidia and other GPU manufacturers, which could include strategic adjustments or new product announcements in response to a potential competitor.

Updates on subsequent generations of the Jalapeño chip, indicating OpenAI’s long‑term hardware roadmap and its impact on the AI ecosystem.

Related guides & quizzes

AI Models ExplainedFuture of AIAI TrainingTest what you know — try a free AI quizLook up an AI term in our glossaryFollow the AI model release tracker
Found this useful?