Nvidia is not the only significant player for major AI investors.
In 2025, some of the world’s largest technology companies are planning substantial capital expenditures. Four major tech giants are projected to collectively spend $320 billion this year, with a significant portion earmarked for constructing and upgrading artificial intelligence (AI) data centers to train and implement generative AI.
Nvidia (NASDAQ: NVDA) is anticipated to benefit considerably from this investment. Major tech firms have emphasized the importance of their partnerships with Nvidia, a leading designer of graphics processing units (GPUs). In addition, statements from Meta Platforms and Alphabet, Google’s parent company, indicate promising prospects for another chipmaker.
Broadcom (NASDAQ: AVGO) may see substantial gains from increased tech spending in 2025 and beyond. Meta and Google have offered insights that could impact Broadcom investors.
Broadcom produces various semiconductors, playing a critical role in AI data centers by manufacturing chips for network switches that facilitate efficient data flow among servers. These switches ensure Nvidia’s chips perform optimally in Meta’s and Google’s data centers. Moreover, Broadcom develops chips utilized in Meta and Google’s custom solutions. Meta’s custom AI chip, the Meta Training and Inference Accelerator (MTIA), incorporates Broadcom’s technology.
Since 2015, Broadcom has collaborated with Google to develop the Tensor Processing Unit (TPU). Both Meta and Google expressed favorable opinions about their custom chips during their fourth-quarter earnings calls.
Meta’s chief financial officer, Susan Li, informed analysts in a call that MTIA deployment for ranking and recommendation tasks related to ads and organic content began in 2024, with further expansion planned for 2025. Over time, Meta aims to transition more of its AI data centers from Nvidia’s general GPUs to its chips. Li mentioned plans to extend MTIA’s capabilities to support core AI training workloads and various generative AI applications.
Alphabet CEO Sundar Pichai highlighted the positive reception of the company’s sixth-generation TPU, Trillium, in the fourth quarter. The newest TPUs, made available in December, are expected to see a significant increase in usage in 2025.
Custom AI accelerator chips, which are generally more efficient than general-purpose GPUs for numerous AI tasks, are increasingly important as large tech firms scale their data centers. Economizing on efficiencies can significantly affect financial results, making control over the technology stack and expanding custom chip applications crucial as capital expenditure budgets grow.
Broadcom is working with companies such as ByteDance, TikTok’s parent company, to develop custom AI accelerators. In its fourth-quarter earnings call, Broadcom’s management announced partnerships with two new customers for next-generation chips, rumored to be Apple and OpenAI.
Apple is known for utilizing Broadcom’s chips, having previously used Google’s TPUs for its Apple Intelligence system. If Apple seeks more control over its AI training and development, it could quickly become one of Broadcom’s significant clients.
Broadcom anticipates that the market for custom AI accelerators might grow to $60 billion to $90 billion in the next three years. The company accounted for about 70% of this market last year, generating approximately $12 billion in sales. This growth might nearly double Broadcom’s revenue if it maintains its market share, considering its total 2024 sales were $51.6 billion.
Despite its promising future, Broadcom’s valuation is a concern. Its stock currently trades at 37 times forward earnings. Although strong earnings growth is projected, this valuation is challenging to justify. A valuation closer to 30 times the 2025 earnings may make it more attractive to investors, especially with expected earnings growth exceeding 30% this year. Alternatively, faster-than-expected adoption of Broadcom’s custom chips by Meta and Google could justify this valuation.
Investors interested in AI’s future may want to monitor Broadcom, though the stock is currently deemed too costly for a recommendation.
