OpenAI and Broadcom Unveil Custom 'Jalapeño' AI Chip for LLM Inference
OpenAI and Broadcom announced their first custom AI chip, 'Jalapeño,' an ASIC optimized for large language model inference. Designed in nine months, it promises better performance-per-watt than Nvidia GPUs but is less flexible and cheaper. Initial deployment is targeted for late 2026, with full production by 2028. The partnership aims to reduce OpenAI's reliance on external suppliers and lower token costs, though Broadcom shares have fallen due to market concerns.
Cross-source coverage
Wire timeline
Broadcom (AVGO) The Best High Growth Semiconductor to Buy Now?
Broadcom Inc. (NASDAQ: AVGO) is highlighted as a high-growth semiconductor stock following a Reuters report on June 24, 2026, announcing that OpenAI, in partnership with Broadcom, released its first custom AI chip named 'Jalapeño.' Designed specifically for inference—processing AI model data and responding to users—the chip is said to be competitive with Nvidia's Blackwell GPUs and Google's TPUs. Broadcom CEO Hock Tan confirmed the chip's performance, which is already running in labs using OpenAI's GPT-5.3-Codex-Spark model at target power and performance levels. The chip took roughly nine months to design with AI-assisted engineering, and will be manufactured by TSMC. Celestica will build the server systems. OpenAI plans to deploy Jalapeño by the end of 2026.
Yahoo FinanceBroadcom (AVGO) The Best High Growth Semiconductor to Buy Now?
On June 24, 2026, Reuters reported that OpenAI, in partnership with Broadcom Inc. (NASDAQ:AVGO), released its first custom AI chip named Jalapeño, designed specifically for inference. Broadcom's CEO Hock Tan stated the chip is competitive with Nvidia's Blackwell GPUs and Google's tensor processing units. OpenAI plans to deploy Jalapeño by the end of 2026, with the chip already running in labs at target power and performance levels using its GPT-5.3-Codex-Spark model. The design took approximately nine months, aided by AI-assisted engineering, and will be manufactured by TSMC. Celestica will build the server systems housing the chips. The article positions Broadcom as a high-growth semiconductor stock to consider.
Yahoo FinanceBroadcom and OpenAI Unveil Jalapeño AI Chip; Implications for AVGO Stock
On June 24, 2026, Broadcom (AVGO) and OpenAI jointly unveiled Jalapeño, OpenAI’s first custom-built AI chip, designed to power next-generation AI models. The chip is the first in a multi-generation computing platform focused on improving speed, reliability, and cost-efficiency for AI deployment at scale. OpenAI designed the chip based on its AI model knowledge, while Broadcom and Celestica handled chip manufacturing, system integration, networking, and production infrastructure. Early engineering samples are already running workloads including GPT-5.3-Codex-Spark, showing better performance and lower power consumption than leading chips. The partnership strengthens Broadcom’s position as a key AI infrastructure supplier, expanding beyond networking into custom AI processors. AVGO stock has surged 133% in two years but retreated 24.8% from its all-time high amid competitive and sector headwinds. The company now has a market cap of ~$1.80 trillion.
Yahoo FinanceOpenAI and Broadcom Launch Custom AI Chip Jalapeño; Should Nvidia Investors Worry?
On June 24, 2026, OpenAI and Broadcom announced Jalapeño, OpenAI's first custom AI chip designed for inference, marking a strategic move to reduce reliance on Nvidia GPUs. Broadcom's custom-silicon business is surging, with $10.8 billion in AI semiconductor revenue (up 143% YoY) and a $100 billion revenue target for FY2027. The article notes that while this poses a real threat to Nvidia's pricing power over time, Nvidia remains dominant with $81.6 billion in quarterly revenue and an unmatched software ecosystem (CUDA). The analysis concludes that Nvidia's scale and flexibility still give it a significant edge, but investors should monitor the trend toward custom chips.
Yahoo FinanceOpenAI and Broadcom Launch Custom AI Chip; Nvidia Weighs $100B Deal That Fell Through
OpenAI and Broadcom announced the joint development of a custom AI chip called 'Jalapeño,' designed for inference in large-language models like ChatGPT. This marks OpenAI's move to build its own chip stack to reduce reliance on Nvidia. The announcement follows news that Nvidia had previously explored a $100 billion megadeal to invest in OpenAI in exchange for chip usage, but the deal was never finalized. OpenAI CEO Sam Altman downplayed tensions, affirming a continued partnership with Nvidia. The chip is part of OpenAI's plan to deploy gigawatt-scale data centers in partnership with Microsoft and Broadcom, with Broadcom CEO Hock Tan emphasizing its role in sustaining AI infrastructure through 2026.
Yahoo FinanceOpenAI Built Custom AI Inference Chip Jalapeño with Broadcom to Reduce Reliance on Nvidia
OpenAI has unveiled its first custom AI chip, named Jalapeño, designed in-house and manufactured with Broadcom. The chip is specifically built for inference—the process of running AI models after training—which represents the fastest-growing and most recurring cost in AI. According to reports, Jalapeño showed better performance per watt than current state-of-the-art processors in early testing. OpenAI went from early schematics to fabrication-ready in nine months, a cycle that normally takes years, partially by using its own models to accelerate design. The move is part of a broader trend as major tech companies like Amazon, Google, Microsoft, and Meta develop their own silicon to reduce dependence on Nvidia, which controls roughly 90% of the AI chip market. Broadcom's CEO stated that companies wanting to lead in AI cannot rely on third-party GPUs for such a key part of their operations. The chip targets the inference workload, where Nvidia's dominance is being challenged as AI usage scales to billions of daily interactions.
Yahoo FinanceOpenAI unveils custom AI chip Jalapeno, co-developed with Broadcom
OpenAI has unveiled its first custom-designed computer chip, named Jalapeno, built to run AI models like ChatGPT faster and more efficiently. The chip, co-developed with Broadcom, is specifically designed for AI inference—the process of generating answers from trained models—rather than training. Early testing indicates substantially better performance per watt compared to current state-of-the-art chips, according to OpenAI. The company said its own AI models helped design the chip, accelerating development. Jalapeno is compatible with a broad range of AI models, not just OpenAI's, and will be deployed at data centers operated by Microsoft and other partners starting in 2026. The move aims to reduce OpenAI's dependence on Nvidia, following similar custom chip strategies by Google, Amazon, and Microsoft. Broadcom CEO Hock Tan called the chip 'just the beginning,' with plans for successive generations.
The Hindu: Latest News today from India and the World, Breaking news, Top Headlines and Trending News Videos.OpenAI Spices Up the Tech Race with Custom ‘Jalapeño’ AI Chip
OpenAI has developed its own custom artificial intelligence chip, dubbed 'Jalapeño,' entering the competitive field of custom AI accelerators. The move places OpenAI alongside tech giants like Meta Platforms and Alphabet, who have also built proprietary AI chips to reduce reliance on external suppliers such as Nvidia. The article, published on June 24, 2026, by Barron's and sourced via Yahoo Finance, highlights the increasing trend of major AI companies designing in-house hardware to optimize performance and cost for their specific AI workloads. The chip's name suggests a spicy addition to the technology race, signaling OpenAI's strategic pivot toward vertical integration in AI infrastructure.
Yahoo FinanceBroadcom and OpenAI unveil custom-built Jalapeño inference processor — OpenAI's first chip is a massive reticle-sized ASIC built in an ultra-fast nine-month development cycle
OpenAI and Broadcom have announced Jalapeño, a custom-built inference ASIC designed for large language models and future agentic AI workloads. The processor, OpenAI's first chip, is a reticle-sized (~840 mm2) compute chiplet surrounded by six HBM3/4 memory modules. It was developed in a nine-month cycle based on OpenAI's understanding of LLM behavior, targeting inference bottlenecks such as data movement, compute-memory balance, and networking efficiency. The companies claim significantly higher performance-per-watt than current state-of-the-art hardware like AMD's Instinct MI350 and Nvidia's Blackwell, though no specific benchmarks or memory configurations were disclosed. Engineering samples are already running in the lab at target clock speeds and power, reportedly executing workloads including GPT-5.3-Codex-Spark. OpenAI stresses the processor is a purpose-built inference chip, not a repurposed training accelerator.
Latest from Tom's HardwareBroadcom and OpenAI Unveil Custom Jalapeño Inference Processor for LLMs
OpenAI and Broadcom have jointly introduced Jalapeño, a custom-built inference ASIC designed specifically for large language models and future agentic AI workloads. The chip is claimed to deliver substantially better performance-per-watt than current state-of-the-art hardware like AMD's Instinct MI350-series and Nvidia's Blackwell-based accelerators, though no hard numbers or benchmarks were disclosed. Jalapeño is a purpose-built inference processor, not a repurposed training accelerator, and features a large compute chiplet surrounded by six HBM modules. The companies state that engineering samples are already operating in the lab at target clock speed and power, running workloads such as GPT-5.3-Codex-Spark. The compute chiplet is estimated to be around 840 mm², near the reticle size limit of EUV lithography. OpenAI considers this the first generation of its inference hardware and calls the project strategically important.
Latest from Tom's HardwareBroadcom and OpenAI unveil Jalapeno custom AI chip to boost stock and lower token costs
Broadcom shares have fallen nearly 20% since early June following a sub-optimal quarter. On June 24, 2026, the company unveiled Jalapeno, a custom AI accelerator chip co-designed with OpenAI, aiming to lower token costs by improving performance per watt. The new chip is the first joint project since the Broadcom-OpenAI custom chip deal was announced eight months ago. While Broadcom CEO Hock Tan expects initial deployment later this year, scaling in 2027, and full production in 2028, OpenAI President Greg Brockman stated the real problem is overall lack of computing power, not cost. The announcement comes amid news that Alphabet and ByteDance are exploring alternate chip partners, though analysts view this as supply chain diversification rather than replacement. The partnership is seen as complementary to Nvidia GPUs, not a replacement. The article concludes the news may improve sentiment but does not yet warrant upgrading the stock to a buy rating.
US Top News and AnalysisOpenAI and Broadcom Unveil Custom AI Chip for Faster, Cheaper Model Runs
OpenAI has announced its first custom artificial intelligence chip, developed in partnership with Broadcom. The finalized chips are designed to run AI models faster and more cheaply, and are slated for integration into large data centers later this year. OpenAI plans to invest tens of billions of dollars in Broadcom chips, signaling a major strategic shift toward custom hardware to reduce reliance on external suppliers and optimize performance for its AI workloads. The announcement was reported by The Business Times Singapore on June 24, 2026.
The Business TimesOpenAI and Broadcom Unveil First Custom AI Chip 'Jalapeño'
OpenAI and Broadcom have announced their first joint custom chip, named 'Jalapeño', an ASIC-based AI accelerator designed for inference workloads in OpenAI's models. The chip, developed in nine months, is part of OpenAI's strategy to reduce reliance on Nvidia GPUs by building its own silicon. Broadcom will manufacture the chips, with initial deployment targeted by end of 2026. OpenAI has also secured chip agreements with AWS (Trainium), AMD, and Cerebras to meet surging demand. The chip is described as less flexible but cheaper than Nvidia GPUs, optimized for specific AI tasks.
US Top News and AnalysisOpenAI and Broadcom Unveil Custom AI Chip 'Jalapeño' for LLM Inference
OpenAI and Broadcom have announced Jalapeño, a custom-designed AI chip specifically optimized for large language model (LLM) inference. The chip aims to significantly enhance performance, energy efficiency, and scalability across AI systems. This collaboration marks a strategic move by OpenAI to reduce reliance on external chip suppliers and tailor hardware to its evolving AI workloads. Jalapeño is expected to accelerate inference speeds and lower operational costs for deploying models like GPT. The announcement underscores a growing trend among AI leaders to develop proprietary silicon to meet the unique demands of generative AI.
OpenAI NewsOpenAI and Broadcom Unveil Custom AI Chip 'Jalapeño' for LLM Inference
OpenAI and Broadcom have announced the introduction of Jalapeño, a custom-designed artificial intelligence chip specifically optimized for large language model (LLM) inference. The chip aims to significantly enhance performance, energy efficiency, and scalability across OpenAI's AI systems. This strategic collaboration marks a major step for OpenAI in reducing reliance on external chip suppliers and tailoring hardware to its specific algorithmic needs. The announcement underscores a growing trend among AI leaders to develop proprietary silicon to handle the immense computational demands of advanced AI models, potentially lowering costs and accelerating deployment capabilities.
OpenAI NewsOpenAI and Broadcom Unveil Jalapeño Custom AI Chip for LLM Inference
OpenAI and Broadcom have announced Jalapeño, a custom AI accelerator designed specifically for large language model (LLM) inference. The chip was developed from scratch in nine months, leveraging OpenAI's deep understanding of LLM fundamentals and Broadcom's silicon implementation expertise. Early testing indicates substantially better performance per watt than current state-of-the-art accelerators, with architecture optimized to reduce data movement and balance compute, memory, and networking resources. Jalapeño is part of OpenAI's full-stack infrastructure strategy, aiming to make AI faster, more reliable, and more affordable. The chip is designed to work with current and future LLMs across the industry, including OpenAI's own models like GPT-5.3-Codex-Spark. Broadcom and OpenAI plan a multi-generation compute platform, with deployment at gigawatt-scale data centers with Microsoft and other partners beginning in 2026. The announcement marks a significant step in OpenAI's vertical integration from products and models to custom hardware.
OpenAI News