Broadcom: The Rising AI Chip Giant
In the past, people might have only recognized NVIDIA, AMD, and TSMC, but now another low-profile yet powerful AI chip giant, Broadcom (AVGO), is making a strong debut. Not only has its performance skyrocketed, but it also briefly surpassed a market capitalization of $1 trillion by the end of 2024, overtaking Tesla and TSMC to rank ninth among the world’s top ten companies by market capitalization.
Why Broadcom Deserves Attention
Core Design and Supply Capability of AI Chips
Hardware and software provide a stable revenue source; even amidst chip market fluctuations, it can balance risks through enterprise software.
Major collaborators include technology giants like Google, Meta, and AWS.
Business Model Breakdown: A Dual Approach of Hardware and Software
Broadcom’s business model relies on two main segments:
- Semi-conductor Business (58%): Provides high-end network switching chips, customized AI chips (ASIC), storage controllers, base station chips, etc.
- Infrastructure Software (42%): Enhances stickiness by acquiring other service providers to offer virtualization, network security, IT automation tools, etc.
This allows Broadcom to survive simultaneously in three major fields:
- Cloud computing infrastructure
- AI training chip supply chain
- Enterprise software services
It is one of the few companies that can earn from AI hardware all the way to software services.
$61 Billion Acquisition of Virtualization Software Company VMware, Entering the Software Market
Broadcom originally started as a high-end chip company, known as Avago Technologies. In 2016, Avago acquired the original Broadcom and retained the name, gradually expanding its product line to include:
- Data centers
- Broadband
- Wireless networking chips
- Storage
- Industrial applications
Subsequently, in 2023, Broadcom acquired cloud computing and hardware virtualization service provider VMware for $61 billion, officially entering the infrastructure software market. After integrating VMware, what did Broadcom gain?
- Control over technology regarding “how AI systems are deployed to cloud servers.”
- VMware’s technology allows enterprises to install, operate, and manage AI models in a cloud environment more easily without needing to rewrite much of the underlying architecture.
Seizing the AI Gold Rush: From Data Centers to Customized Chips
With the rise of generative AI, tech giants are rushing to build AI servers and actively train AI models, leading to a surge in chip demand. In response, NVIDIA has dominated the mass GPU market, while Broadcom has taken a different route, focusing on “customized AI chips” (ASIC) for cloud giants. For instance, Google’s Tensor Processing Unit (TPU) is designed by Broadcom, and Meta has also partnered with them to develop internal AI chips.
This has enabled Broadcom’s AI semiconductor revenue to reach $12.2 billion in 2024, a year-on-year increase of 220%, with AI business accounting for 41% of the company’s semiconductor revenue, making it another AI chip giant.
The diagram provided by Broadcom shows the Google TPU breakdown, where the main chips are supplied by Broadcom.
Record High Revenue and Market Capitalization, Surpassing $1.15 Trillion
Broadcom’s total revenue reached $51.57 billion in the 2024 fiscal year, with semiconductor contributions still accounting for the majority (58%), while software (42%) is also growing rapidly. Even more astonishing, at the end of 2024, Broadcom’s market capitalization first exceeded $1 trillion and reached $1.15 trillion in 2025, alongside Apple, Microsoft, and NVIDIA, moving into the top ten global companies by market capitalization.
The chart shows Broadcom setting a record high market capitalization of $1.15 trillion in January 2025.
Future Outlook: AI Chip Revenue Expected to Reach $90 Billion
According to Broadcom CEO Hock Tan in the December 2024 earnings call, it is estimated that by 2027, the market size Broadcom can participate in, sell products, and earn profits from the entire AI chip market will grow to $60 billion to $90 billion. This forecast reflects the strong growth in AI chip demand and drove Broadcom’s stock price to rise 21% on that day, surpassing a market capitalization of $1 trillion. The acquisition of VMware is aimed at “retaining enterprise customers” and strengthening the foundation for AI applications.
Will Broadcom Be NVIDIA’s Strongest Competitor?
From a technical perspective, not necessarily, because:
- Broadcom focuses on customized AI ASIC chips, binding itself to tech giants like Google, Meta, TikTok, and AWS.
- NVIDIA emphasizes high-end chips centered around GPUs that are accessible to all.
Although NVIDIA still dominates the AI chip industry in terms of revenue and influence, Broadcom is gradually establishing itself in the core businesses of cloud customers. As major platforms increasingly invest in self-developed AI chips, Broadcom’s role will become more critical.
Risk Warning
Investing in cryptocurrencies carries high risks, and prices may fluctuate dramatically, potentially resulting in the loss of your entire principal. Please assess the risks carefully.