By Yang Jianyong

In the second quarter of fiscal year 2026, Broadcom posted a revenue of $22.2 billion, reflecting a significant 48% year-over-year surge. Notably, revenue from AI chips skyrocketed to $10.8 billion, marking a staggering 143% year-over-year increase. However, this impressive revenue growth was not enough to stave off a major market sell-off, causing Broadcom's market value to plummet by $285.6 billion (approximately RMB 1.93 trillion) overnight.

Broadcom is widely regarded as Nvidia's toughest competitor. It's crucial to understand that in the global AI data center market, Nvidia commands an estimated market share of over 80%, and possibly even up to 90%. This dominant position underscores the heavy reliance on Nvidia's high-performance chips for training models across diverse industries, painting a picture of a near-monopoly.

However, amidst a backdrop where tech behemoths are seeking to diversify their chip supply and ramp up their in-house chip development efforts, they are increasingly turning to customized chips. Examples include Google's TPU chips, Amazon Web Services' Tranium chips, and Alibaba's Pingtouge Zhenwu chips. These bespoke solutions not only cater more effectively to their specific needs but also reduce their dependency on Nvidia.

In this evolving landscape, there's a burgeoning demand for customized AI chips. Broadcom, as a leading player in the custom AI chip arena, is reshaping the global AI chip industry through its customization strategy, developing tailored AI chips for industry giants like Google, Meta, OpenAI, and Anthropic.

Customized AI chips are also emerging as a key driver of Broadcom's revenue growth. In its latest earnings report, Broadcom CEO Hock Tan highlighted that the accelerated growth in AI chip revenue propelled the company to record-breaking levels in terms of revenue, operating profit, and free cash flow in the second quarter. AI chip revenue surpassed expectations, primarily due to the escalating demand for custom AI chips and AI networking solutions.

Tan emphasized that this growth trajectory is set to continue, with the AI semiconductor business projected to reach $16 billion in the third quarter, representing a 200% year-over-year increase. Meanwhile, management reaffirmed that the AI semiconductor revenue guidance for fiscal year 2027 will exceed $100 billion. Nevertheless, Broadcom did not raise its full-year AI revenue forecast, which still falls short of institutional projections.

Analysts from JPMorgan anticipate Broadcom's AI revenue to hit $120 billion by 2027, while Jefferies analysts go even further, predicting it could exceed $200 billion.

Ultimately, with major tech companies continuing to ramp up their capital expenditures on AI infrastructure, the demand for chip-level computing power for large models is also being unleashed, sustaining a robust demand for AI chips.

According to the latest forecast data from Goldman Sachs, the AI spending of just four tech giants—Meta, Microsoft, Amazon, and Google—is projected to reach a staggering $5.3 trillion by the end of 2030.

This global trend of expanding capital expenditures signals that these cloud giants possess substantial cash reserves and significant free cash flow to bolster their competitiveness in the AI market. This, in turn, is propelling the global semiconductor market towards a historic trillion-dollar milestone and positively influencing the performance of AI computing chip companies.

Broadcom boasts a formidable competitive edge in the AI ASIC/XPU market, as evidenced by its impressive financial results. These results not only showcase the strength of customized AI chips but also hint at the potential to disrupt Nvidia's dominance in the AI chip arena. Meanwhile, with the gradual release of market demand for AI, ASIC represents the next high-growth sector, which is why Broadcom's market value has the potential to join the elite $2 trillion club.

Yang Jianyong, a contributor to Forbes China, expresses views that are solely his own. He is committed to providing in-depth interpretations of cutting-edge technologies such as AI large models, artificial intelligence, Internet of Things, cloud computing, and smart home appliances.