© Justin Sullivan / Getty Images
Nvidia (NASDAQ:NVDA | NVDA Price Prediction) delivered another blockbuster quarter, reporting fiscal fourth-quarter revenue and earnings that easily trounced Wall Street’s expectations as demand for its advanced artificial intelligence (AI) accelerators remains white hot. With enterprise and hyperscaler demand for AI infrastructure showing no signs of slowing, the chipmaker’s fundamentals remain rock-solid.
Yet Nvidia’s stock is tumbling 5% in morning trading today as AI fatigue, sky-high valuations potentially limiting further upside opportunity, and concerns about heavy customer concentration rippled across the sector. Broadcom (NASDAQ:AVGO) shares were down 6.5% in sympathy, even as the company made its own landmark announcement this morning.
Gloom Amid the Boom
Nvidia’s results are forcing investors to reassess the competitive landscape for every AI-chip supplier, including Broadcom. The message from the market appears to be: if even Nvidia — the undisputed GPU king with an unmatched software moat in CUDA — can’t excite investors after crushing numbers, what chance do the runners-up have?
Skeptics worry that Broadcom’s custom-ASIC business could face headwinds. Hyperscalers are spending tens of billions on Nvidia GPUs; any slowdown in overall AI capex would hit everyone. Some fear custom silicon ramps may prove slower or more expensive than hoped, and that Broadcom’s heavy exposure to the same Big Tech customers as Nvidia creates correlated risk. Valuation concerns that hammered Nvidia are now being applied across the board.
Broadcom investors, however, should not be alarmed. In fact, today’s news from Broadcom is precisely why the sell-off represents a buying opportunity.
The Spark That Ignites Broadcom’s Rocket
Broadcom announced it has begun shipping the industry’s first 2 nanometer (nm) custom compute system-on-a-chip (SoC), built on its proprietary 3.5D eXtreme Dimension System in Package (XDSiP) platform. This is no incremental upgrade: The chip powers Fujitsu’s Fujitsu-Monaka high-performance, low-power processor for AI and high-performance compute (HPC), with volume shipments of additional customer XPUs slated for the second half of 2026.
Developed using Taiwan Semiconductor Manufacturing‘s (NYSE:TSM) 2nm process node, the chip delivers roughly 10% to 15% higher performance or 25% to 30% lower power versus 3nm at the same performance. When combined with Broadcom’s 3.5D Face-to-Face stacking, the architecture allows independent scaling of compute, memory, and I/O dies in an ultra-dense package. The result is unprecedented compute density, dramatically better energy efficiency, and lower latency — exactly what hyperscalers need to keep scaling AI clusters without collapsing power grids or budgets.
This milestone cements Broadcom’s dominance in the custom-silicon market. Analysts at Counterpoint Research project Broadcom will hold approximately 60% share of AI server compute ASICs by 2027. The company already designs chips for Google’s TPUs, Meta Platforms‘ (NASDAQ:META) MTIA, and reportedly others, including OpenAI and Anthropic. Its AI-related backlog stands at $73 billion, providing multi-year visibility.
Nvidia — Not Broadcom — Should Worry
Custom ASICs are among Broadcom’s highest-margin products and sit alongside its dominant AI networking portfolio (Tomahawk and Jericho switches), creating a full-stack offering few rivals can match.
Far from being threatened by Nvidia’s strength, Broadcom’s 2nm achievement actually puts pressure back on Nvidia. Hyperscalers are aggressively pursuing “build versus buy” strategies because custom silicon delivers superior performance-per-watt and lower total cost of ownership for their specific model architectures — especially in inference, where efficiency matters most. Every dollar spent on a Broadcom XPU is a dollar not spent on a Nvidia GPU. As power consumption becomes the binding constraint in data centers, the efficiency edge of 2nm custom designs becomes decisive.
Broadcom is not trying to replace Nvidia’s GPUs; it is carving out the high-margin, high-volume portion of the market where specialization wins. That strategy is working. AI semiconductor revenue is expected to continue doubling or more in the coming years, and the 2nm platform de-risks the entire roadmap.
Key Takeaway
Nvidia’s results prove it continues to dominate the broad AI GPU market, but they do not diminish Broadcom’s differentiated opportunity. While the Street frets over sector rotation and valuations, Broadcom is executing flawlessly on its strengths: advanced packaging leadership, custom-silicon expertise, and networking integration. The 2nm SoC announcement is not just a technical win — it is a multi-billion-dollar growth catalyst that will drive incremental revenue for years.
At current levels, the 6.5% pullback in Broadcom’s shares looks like a gift. Investors who focus on execution and long-term AI infrastructure tailwinds should view today’s weakness as the entry point they have been waiting for.