Is Nvidia’s five-year run as the undisputed AI stock champion about to be disrupted? Quietly, a different chip titan has joined the $2 trillion club and is now setting the pace in 2026.
Broadcom (AVGO) is no longer just a networking behemoth that touches 99% of global internet traffic. It is morphing into a custom silicon powerhouse with its own road to AI riches — one that does not require going head-to-head with Nvidia’s all-conquering GPUs. That distinction is now translating into real market leadership: Broadcom shares have gained roughly 16% year-to-date and just hit a fresh all-time high, while Nvidia, for all its dominance, has climbed around 7%.
The catalyst that pushed Broadcom into record territory was twofold. On April 21, Vanguard executed share splits across five of its low-cost ETFs. In four of those five funds, Broadcom ranked among the top ten holdings — a quiet reminder of how deeply the stock is already embedded in passive portfolios. The next day, Broadcom rallied on broader market strength and news that its partnership with Alphabet’s Google Cloud is expanding, both in network observability through AppNeta and in the co-design of Google’s next-generation TPU chips, the TPU 8t for training and TPU 8i for inference.
This is not the Nvidia playbook. Nvidia sells general-purpose GPUs that customers purchase and then tune for their own AI workloads. Broadcom, by contrast, builds XPUs — custom processors designed in close collaboration with individual hyperscale clients. These chips are purpose-built from the silicon up for a specific customer’s AI stack. The outcome is a deep, sticky relationship: Meta, Alphabet, and other mega-cap cloud operators are not just buying chips from Broadcom; they are co-engineering their AI infrastructure with the company. Broadcom calls these ties “deep, strategic, and multiyear.”
The numbers attached to that strategy are staggering. Broadcom management has laid out a path to more than $100 billion in annual AI chip revenue by fiscal 2027, and has publicly verified that its supply chain can support that volume. That forecast — backed by confirmed orders and expanding relationships — places Broadcom firmly at the center of the AI buildout, without ever selling a single GPU.
Defensively, Broadcom offers what few AI-focused mega-caps can. Beneath the AI growth engine sits a reliable stream of free cash flow from its non-AI semiconductor and software infrastructure businesses. That fortress-like base has funded consistent stock buybacks and just delivered the company’s 15th straight year of dividend increases. In a market still jittery about the durability of AI spending, that mix of offense and defense is powerful.
Valuation, too, has swung in Broadcom’s favor on a sentiment basis. Both stocks have seen their multiples compress from euphoric highs to more reasonable levels. Today, Nvidia actually trades on a lower forward earnings multiple than Broadcom — yet it is Broadcom catching the momentum bid as money rotates out of extended winners into AI names that still carry a new burst of growth potential.
None of this is a verdict against Nvidia. Its GPUs remain the standard for general AI compute, and quarterly results can still surprise. But stretched over the remainder of 2026, the stock that ends the year atop the Nasdaq may not be the one everyone has been holding for half a decade. Broadcom’s custom-silicon roadmap, its $100 billion revenue target, and its freshly-minted $2 trillion market cap make a credible case that the engine driving portfolio returns in the AI space could already be switching.
