Prediction: Broadcom and Nvidia Will Be 2027's Best Performing AI Stocks

Nvidia expects revenue to grow 70% in 2027, driven primarily by increased AI-hyperscaler spending and its higher-priced Rubin architecture, while Broadcom projects AI semiconductor revenue will double in both 2027 and 2028. The article argues that despite 2026 share gains of about 17% for Nvidia and less than 5% for Broadcom, versus roughly 11% for the S&P 500, their valuations do not reflect the projected AI growth. Using a 30x trailing P/E assumption and consensus earnings forecasts, it estimates more than 100% upside for Nvidia and over 50% for Broadcom by the end of the next fiscal year.

Analysis

The investable distinction is not simply AI exposure but revenue quality. NVDA’s architecture transition can preserve system-level pricing if customers continue buying full racks, networking and software; the key risk is that hyperscalers shift a larger share of incremental workloads to internally designed ASICs, which benefits AVGO but caps NVDA’s accelerator mix and raises the probability of a valuation de-rating even amid earnings growth. AVGO’s upside is more dependent on a small number of bespoke-program ramps, where customer concentration, tape-out timing and lower custom-silicon gross margins can make revenue growth materially less valuable than equivalent NVDA growth.

Over the next 1-3 months, year-end positioning can support both names, but the durable catalyst is evidence that 2027 capex is incremental rather than merely a reallocation from general compute, networking, memory or data-center power budgets. Watch META, MSFT, GOOGL and AMZN capex guidance, as well as lead times at TSMC: accelerated CoWoS capacity additions would validate volume but could ultimately loosen supply and weaken NVDA’s scarcity premium in 6-18 months. Second-order beneficiaries of sustained deployment are TSM (advanced packaging), ANET (scale-out networking) and VRT/ETN (power and thermal infrastructure); MU benefits only if HBM pricing remains tight rather than if supply catches up.

Consensus appears too willing to capitalize multiyear management targets at a static earnings multiple. The more relevant 2027 question is whether incremental AI revenue carries stable gross margin and whether depreciation from hyperscaler build-outs begins to constrain 2028 purchasing; a single large customer program delay would disproportionately challenge AVGO, while a Rubin qualification or rack-integration delay would challenge NVDA. Falsify the constructive view if the next two hyperscaler reporting cycles show aggregate capex growth decelerating below current run-rate expectations, NVDA data-center gross margin falls materially without an offsetting revenue beat, or AVGO discloses AI backlog conversion delays.

Key Decisions for Investors

• Initiate a 3-6 month long NVDA / short AVGO pair, sized beta-neutral, if both remain bid into year-end: NVDA has broader customer diversification and higher exposure to full-system monetization, while AVGO carries greater custom-ASIC program and margin-mix risk. Exit if AVGO reports backlog conversion ahead of schedule with stable segment margins, or if NVDA’s Rubin ramp slips.
• Maintain a core long TSM through the next two earnings cycles as the lower-valuation supply-chain expression of AI compute growth; use a 10-12% downside stop tied to evidence of advanced-packaging utilization easing. Risk/reward improves if hyperscaler capex guidance is raised before TSM’s capacity commentary catches up.
• Add ANET and VRT on post-earnings pullbacks rather than chase semiconductor beta: networking and power build-outs monetize cluster deployment regardless of whether the accelerator is NVIDIA-designed or a Broadcom-enabled custom ASIC. Review the thesis if hyperscaler capex shifts toward land/building spend while network-port and power-backlog commentary weakens.
• Do not underwrite the article’s implied upside solely from a multiple reset. Set alerts for aggregate AMZN/MSFT/GOOGL/META capex guidance and TSM advanced-packaging commentary; absent upward revisions in either, treat rallies in NVDA and AVGO as positioning-driven rather than a new earnings-estimate cycle.