Broadcom (NASDAQ: AVGO) continues to make significant strides in the tech industry, with recent announcements of extended partnerships and new developments. Last week, the company sealed a deal with Meta Platforms (NASDAQ: META) to collaborate on the creation of custom AI chips. This agreement, which spans until 2029, involves an initial commitment of 1 gigawatt worth of chips, marking the beginning of a multi-generation custom chip roadmap. As part of this deal, Broadcom CEO Hock Tan will transition from Meta’s board to an advisory role to assist in shaping the custom chip roadmap.
Furthermore, Meta recently unveiled four generations of custom MTIA (Meta Training and Inference Accelerator) chips, with the MTIA 300 already in use for ranking and recommendation training. The upcoming MTIA 400, 450, and 500 models are designed specifically for handling AI workloads, focusing on inference capabilities.
In addition to its collaboration with Meta, Broadcom has extended its partnership with Alphabet to aid in the development of future iterations of Tensor Processing Units (TPUs). This partnership also includes a deal to supply Anthropic with an additional 3.5 gigawatts worth of chips starting in 2027. With $21 billion in TPU orders from Anthropic for this year alone, Broadcom’s foothold in the AI chip market is solidifying.
Prior to these developments, Broadcom had projected delivering $100 billion worth of custom AI chips to customers by fiscal 2027. This is a significant increase from the company’s total AI revenue of around $20 billion in fiscal 2025. With these extended partnerships and deals, Broadcom is poised for continued growth in the years ahead.
Broadcom’s custom chip business plays a crucial role in supporting its data center networking segment. The company is a key player in the networking space, with its Tomahawk Ethernet solution leading in switching technology. As demand for AI chip clusters rises, Broadcom’s networking business is experiencing substantial growth.
Hyperscalers, who own large data centers, are increasingly exploring alternatives to Nvidia’s high-priced GPUs. Custom AI chips are emerging as a preferred option for hyperscalers, especially for inference tasks. Custom AI ASICs offer performance and cost advantages, being purpose-built and energy-efficient. With Broadcom’s expertise in this area, more companies are turning to the company for assistance in developing custom AI solutions.
In conclusion, Broadcom’s recent partnerships and developments in the AI chip space position the company for significant growth in the tech industry. With a strong foundation in custom chip technology and a growing demand for AI solutions, Broadcom is well-positioned for success in the coming years.
