Broadcom just delivered a knockout punch to earnings expectations, with AI revenue rocketing 63% to $5.2 billion in Q3 and guidance that suggests the chip giant is positioning itself as Nvidia's most credible challenger. The results sent a clear signal: the AI infrastructure boom has multiple winners, and Broadcom is claiming its throne.
Broadcom just proved that Nvidia doesn't get to have all the AI party to itself. The semiconductor powerhouse smashed through Q3 expectations with a staggering 63% jump in AI revenue to $5.2 billion, while total revenue climbed 22% year-over-year to $15.96 billion according to fiscal third-quarter earnings released Thursday. The performance handily beat the $15.83 billion Wall Street consensus and delivered adjusted earnings of $1.69 per share versus the $1.65 expected. What's grabbing investor attention isn't just the beat – it's the forward momentum. CEO Hock Tan guided Q4 revenue to $17.4 billion, well above the $17.02 billion analysts were modeling, and projected AI revenue alone will reach $6.2 billion next quarter. That trajectory puts Broadcom on track to generate roughly $20 billion in annual AI revenue, a figure that would make it the second-largest AI chip player behind only Nvidia's estimated $60+ billion run rate. The stock market is taking notice. Broadcom shares have surged 32% year-to-date and nearly doubled over the past 12 months, pushing the company's market capitalization past $1.4 trillion. That puts it in rarified air alongside Apple, Microsoft, and Nvidia in the trillion-dollar club, with investors betting that custom AI accelerators represent a genuine alternative to Nvidia's one-size-fits-all approach. Broadcom's secret weapon lies in its custom chip development partnerships with hyperscale cloud providers. The company designs bespoke AI accelerators tailored specifically for Google, Meta, and other cloud giants' unique workloads, rather than selling general-purpose GPUs. Tan revealed in March that Broadcom was developing next-generation AI chips with three major cloud customers, positioning these custom solutions as potentially more efficient and cost-effective than off-the-shelf alternatives. The strategy is paying dividends across Broadcom's portfolio. Semiconductor solutions revenue jumped 57% to $9.17 billion, driven primarily by AI accelerator demand, while the company's infrastructure software business – anchored by its $69 billion VMware acquisition – grew 43% to $6.79 billion. The VMware integration has been particularly crucial, providing the networking and virtualization software that ties thousands of AI chips together in massive data center deployments. Industry observers are watching whether Broadcom's custom approach can scale to challenge Nvidia's ecosystem dominance. While Nvidia's CUDA software platform and broad developer support have created powerful switching costs, hyperscale customers are increasingly interested in reducing their dependence on a single supplier. Amazon's homegrown Trainium chips, Google's TPU processors, and now Broadcom's custom solutions represent a coordinated effort to break GPU hegemony. The timing couldn't be better for Broadcom's AI ambitions. Enterprise AI spending is accelerating beyond initial ChatGPT experiments toward production deployments that require massive computational infrastructure. Tan attributed the company's Q3 performance specifically to custom AI accelerators, networking components, and VMware software – exactly the stack needed for large-scale AI implementation. Looking ahead, Broadcom faces the challenge of maintaining this breakneck growth pace while competing against Nvidia's entrenched position and emerging rivals like and Intel. The company's $6.2 billion Q4 AI revenue guidance suggests confidence that current hyperscale partnerships will continue expanding, but sustaining 60%+ growth rates will require either deeper penetration with existing customers or landing additional major cloud providers.
