3 Artificial Intelligence (AI) Stocks That Are Expected to Grow at Faster Rates Than Nvidia Over the Next Couple of Years, According to Analysts

The article argues Nvidia’s growth should slow to a 26.2% CAGR through 2028, while Broadcom (35.6%), AMD (35.2%), and Marvell (30.3%) are projected to grow faster on AI/custom-chip demand. It highlights Broadcom’s $1.9 trillion market cap and 78 P/E, AMD’s MI450 launch and OpenAI/Meta partnerships, and Marvell’s $2 billion Nvidia investment and Google win. Overall, the piece is a comparative valuation/growth analysis rather than a new company-specific catalyst.

Analysis

The market is starting to re-rate AI semis from a single-winner narrative to a platform-picks-and-shovels framework. That matters because custom silicon monetizes a different part of the stack: hyperscalers can preserve AI capex while lowering unit economics, which transfers margin away from merchant GPUs and toward design wins, IP, and packaging-heavy suppliers. The second-order effect is not that NVDA loses absolute growth immediately, but that its share of incremental wallet spend likely narrows as buyers diversify sourcing and push more workload-specific chips into production.

The most interesting setup is actually the spread between growth and valuation. AVGO, AMD, and MRVL all screen as higher-growth than NVDA on a forward CAGR basis, but they are also taking on execution risk at a stage where expectations are now elevated and multiple compression can offset operating beats. AMD has the cleanest near-term catalyst stack because product launch timing can reframe the earnings trajectory over the next 2-3 quarters; MRVL has the best torque to custom-chip design wins, but it is also the most dependent on a handful of large customers and therefore the most exposed to order volatility if one hyperscaler pauses.

Contrarianly, the consensus may be underestimating how sticky NVDA’s ecosystem remains. Even when customers adopt custom silicon, they often keep NVDA as the training backbone and use alternatives only for inference or edge workloads, which means the “share loss” story is more about mix than outright displacement. That suggests the right trade is not a blanket short NVDA, but a relative-value rotation into the names with the highest probability of near-term surprise, while respecting that a broad AI capex slowdown would hit the whole complex first and fastest.

From a timing perspective, the next inflection window is the next two earnings cycles, not the next two weeks. If hyperscaler commentary confirms custom-chip ramp and capex budgets remain intact, this can sustain a 3-6 month rotation trade; if guidance turns more cautious, high-multiple semis will de-rate together and the pair will matter less than gross beta.

Key Decisions for Investors

• Long AMD / short NVDA as a 3-6 month relative-value trade: AMD has the clearest product-cycle catalyst, while NVDA carries the highest risk of multiple compression if investors decide growth is normalizing; stop if NVDA re-accelerates cloud/enterprise demand or AMD launch adoption disappoints.
• Overweight AVGO on pullbacks versus NVDA and MRVL: AVGO has the best combination of custom-silicon monetization and balance-sheet durability, making it the cleaner way to own AI infrastructure if capex stays strong over the next 2 quarters.
• Speculative long MRVL into customer-design-win confirmation, but size small: best upside if Google-style wins broaden, yet the name is most vulnerable to hyperscaler concentration risk; use a 3-6 month horizon and cut if order concentration worsens.
• Use call spreads rather than outright longs in AMD and AVGO: upside remains meaningful, but valuation is already discounting a lot of good news; spreads reduce P/E compression risk if the market starts paying more attention to margins than growth.
• Do not initiate a structural short NVDA here; if anything, express the view as a pair against the faster-accelerating names, because ecosystem stickiness and training demand could keep NVDA resilient even if share shifts at the margin.