Nvidia(NASDAQ: NVDA) has dominated the semiconductor market throughout the artificial intelligence (AI) revolutions. Its graphics processing units (GPUs) are essential infrastructure for companies developing generative AI models.
Naturally, investors are focusing on competitors like Advanced Micro Devices(NASDAQ: AMD) and Intel(NASDAQ: INTC), both of which make their own GPU and CPU architectures. However, I think Nvidia’s biggest competitive threat could come from an entirely different direction.
Missed AI’s “Act 1”? Act 2 Could Be 16x Bigger.Most investors think they missed the AI boat because they didn’t buy Nvidia in 2005. But according to our analysts, we’re only at the end of “Act 1″—the R&D phase. “Act 2” is the global rollout.Continue »
AI hyperscalers are developing their own processors, known as application-specific integrated circuits (ASICs), to reduce costs and optimize performance for particular workloads.
With major customers including Alphabet (NASDAQ: GOOGL) (NASDAQ: GOOG), Meta Platforms (NASDAQ: META), OpenAI, and Anthropic, Broadcom (NASDAQ: AVGO) is emerging as a formidable competitor for future AI infrastructure spending. Here’s why smart investors should pay attention.
Broadcom is helping Nvidia’s biggest customers make their own chips
Nvidia’s GPUs are useful because they can handle a wide variety of accelerated computing tasks. However, that flexibility isn’t always necessary. Custom ASICs are designed for specific applications, potentially offering better performance per dollar of compute and lower energy consumption for certain workloads.
Broadcom is helping design these specialized processors while also supplying networking technology that connects clusters of chips inside data centers. Let’s consider its expanding roster of AI chip customers.
Alphabet is working with Broadcom on its Tensor Processing Units (TPUs), including the seventh-generation Ironwood processor. These chips support Google Cloud and are being deployed for Anthropic. Broadcom expects Anthropic to deploy 5 gigawatts of eighth-generation TPU v8i capacity in 2027, followed by an additional 10 gigawatts in 2028.
Meanwhile, Meta is collaborating with Broadcom on its Meta Training and Inference Accelerator (MTIA) processors. Their partnership extends through 2029, with visibility into 3 gigawatts of deployments through 2028.
Last October, Broadcom and OpenAI announced plans to deploy 5 gigawatts of XPUs by the end of 2028. OpenAI’s first processor, code-named Jalapeño, is expected to support 1.3 gigawatts of deployments in 2027.
