Broadcom Cements AI Role with Google, Samsung HBM Supply Deal
Securing half of Samsung's advanced memory output for Google's AI chips highlights Broadcom's critical position in the custom silicon market.
Broadcom Inc. has made a decisive move to secure its position at the heart of the artificial intelligence hardware boom, locking down a significant portion of Samsung's High-Bandwidth Memory (HBM) production for its custom work on Google's next-generation AI accelerators.
The deal, which dedicates nearly half of the Korean tech giant's HBM output to Broadcom, underscores the chipmaker's indispensable role in the burgeoning market for custom silicon. As hyperscale companies like Google race to develop proprietary chips to power their AI models, Broadcom has emerged as a key enabler, designing the critical application-specific integrated circuits (ASICs) that serve as the brains for these systems, including Google's Tensor Processing Units (TPUs).
This strategic supply agreement is pivotal as HBM has become a major bottleneck in the production of AI hardware. The advanced memory chips are essential for providing the high-speed data access that powerful processors require for training and running large language models. By securing a stable supply, Broadcom not only de-risks its production pipeline for Google's upcoming "Ironwood" TPU but also fortifies its competitive standing against others vying for the limited HBM supply.
Investors have taken note of Broadcom's deepening entrenchment in the AI supply chain. The company's stock has surged over 60% year-to-date, recently touching an all-time high of $403 per share. Its market capitalization now stands at approximately $1.8 trillion, placing it firmly in the top tier of global technology firms. The move comes as Broadcom's AI-related revenue has already seen a 63% year-over-year increase, highlighting the segment's powerful contribution to growth.
The Google partnership is the cornerstone of Broadcom's custom silicon strategy. Projections indicate that shipments of Google's TPUs are expected to quintuple by 2028, providing a long-term revenue stream for Broadcom. This custom design work offers a powerful alternative to the merchant silicon market dominated by Nvidia's GPUs, catering to cloud providers seeking optimized performance and efficiency for their specific AI workloads.
Wall Street has responded with increasing optimism. In recent weeks, analysts from Morgan Stanley, UBS, and Bank of America have reiterated 'Buy' ratings and hiked their price targets for Broadcom. The consensus price target now sits at $407.25, suggesting analysts see further room for growth even after the stock's impressive run. According to a recent report, this confidence is explicitly tied to the surging demand for custom AI chips from cloud and enterprise customers.
Broadcom's AI ambitions are not limited to its work with Google. The company recently confirmed a multi-year partnership with OpenAI to supply custom data center capacity, further diversifying its roster of top-tier AI clients. This strategy of partnering with the biggest names in AI to build bespoke solutions leverages Broadcom's deep engineering expertise and distinguishes it within the crowded semiconductor landscape.
As the company prepares for its upcoming earnings report, the focus will be on the continued growth of its AI business and the successful integration of its recent VMware acquisition, which bolsters its infrastructure software offerings. For now, by securing the critical HBM supply for one of its largest partners, Broadcom has sent a clear signal to the market: it is a foundational and enduring player in the hardware infrastructure that will power the next decade of artificial intelligence.