OpenAI Builds Nvidia-Rival Inference Chip in Nine Months with Broadcom Support
Yahoo Finance ·
OpenAI has unveiled its first proprietary inference chip, Jalapeño, moving from initial architecture to tape-out in just nine months. According to InferenceX data, the new silicon delivers 1.5 to 1.9 times greater work per watt and 1.7 to 3.6 times lower end-to-end latency compared to Nvidia GB200 and GB300 systems. Broadcom Inc. managed the physical implementation and supplied Tomahawk networking, while TSMC manufactured the hardware and Celestica handled system integration. Rated at 700 watts with a sustained power draw of 550 watts or less, Jalapeño focuses strictly on inference rather than training. OpenAI and Broadcom aim to deploy 10 gigawatts of custom accelerators through 2029. Meanwhile, institutional interest remains high, with short interest hovering near 1.20 percent for Broadcom and 1.23 percent for Nvidia.
AI 시장 분석
OpenAI collaborated with Broadcom to develop Jalapeno, an inference chip that outperformed Nvidia's performance in just 9 months. This partnership is expected to benefit Broadcom by proving its custom chip design capabilities in the AI infrastructure market. On the other hand, Nvidia faces a strategic threat of core inference workload defection.
상승 영향
- Semiconductors — Broadcom is expected to accelerate AI semiconductor revenue growth by supporting the physical implementation and networking of OpenAI's custom inference chip Jalapeno.
하락 영향
- Semiconductors — Nvidia suffers a strategic blow as major clients like OpenAI adopt their own inference chips, depriving them of repetitive workloads for premium GPUs.
DYAX 전담 분석
OpenAI's new inference chip Jalapeno records 1.5 to 1.9 times higher AI task efficiency per watt, serving as a powerful driver for Broadcom's earnings growth. Broadcom aims to expand AI semiconductor sales through a 10-gigawatt accelerator deployment plan by 2029.
The bull case is the expansion of Broadcom's custom chip supply and client diversification, while the bear case is margin pressure from high HBM memory costs. Investors should monitor customer concentration and Nvidia's CUDA ecosystem defense as key indicators.
AI가 생성한 분석으로 투자 자문이 아닙니다.
DYAX Investor Sentiment
Bullish (Long) 70% · Bearish (Short) 30%
420 participants
Related News
- Dell's Valuation Multiple Contracts Amid Server Production Bottlenecks
- Anthropic Plans Insider Stock Sale as Part of Historic IPO
- Webull Rallies 8% and Robinhood Gains as Retail Trading Momentum Persists
- Halloween Costume Controversy Threatens Target’s Hard-Won Recovery
- Major Credit Card Firm Distributes $576M in Annual Dividends to Buffett's Berkshire
- Trump Praises Micron's Massive $10B Investment Initiative
