These 2 Artificial Intelligence (AI) Stocks Could Soar by 66% and 46%, According to Wall Street

Micron and Broadcom are pitched as major AI beneficiaries, with Micron’s fiscal Q3 revenue up 346% YoY to $41.46B and adjusted EPS up 1,215% YoY to $25.11 amid an expected memory shortage lasting at least to 2028. Broadcom reported fiscal Q2 revenue up 48% YoY to $22.2B and adjusted EPS up 54% YoY to $2.44, with AI-chip revenue growth projected to exceed 200% YoY in Q3. Wall Street’s implied upside is ~66% for Micron to an average target of $1,515.11 and ~46% for Broadcom to an average target of $526.30, supporting a bullish view despite cyclical-memory risk.

Analysis

The cleaner second-order winner is not just the chip vendors but the hyperscalers buying them: GOOG and META gain if custom silicon plus tighter memory supply lowers cost per inference and improves AI gross margin. That shifts the market from asking whether AI spend grows to how much of that spend is converted into lower unit economics, which is a subtler but more durable source of equity upside. For NVDA, the risk is not demand collapse; it is content-per-dollar compression as customers internalize more of the stack.

MU has the strongest near-term operating leverage, but it is also the most cyclical name here, so the key question is whether this is a 2-3 quarter earnings surprise or a 2-3 year supercycle. The main falsifier is a faster-than-expected supply response: if memory makers or equipment capex step up aggressively, ASPs can roll over well before revenue growth does. That means the thesis is best monitored through pricing commentary and capex plans, not headline revenue growth.

AVGO looks more durable because long-duration customer relationships and ASIC adoption reduce revenue volatility versus merchant silicon. Still, the market may be underestimating substitution risk: if custom chips become the default for inference, NVDA can lose share even while AI capex stays elevated. Consensus may be too focused on absolute AI spend and not enough on mix shift; relative value should matter more than the theme itself.

Key Decisions for Investors

• Long MU on pullbacks for a 3-6 month trade; use the low multiple as a valuation floor, but keep size modest because the thesis is still cyclical. Falsify on any sign of DRAM/NAND ASP deceleration or capex re-acceleration from memory peers.
• Long AVGO as the higher-quality AI compounder into the next hyperscaler capex cycle; better risk/reward than chasing the most expensive AI beneficiaries. Best held 6-12 months, with thesis broken if AI growth guidance stalls for two consecutive quarters.
• Relative value: long MU / short NVDA small size as a hedgeable way to express 'memory scarcity is underappreciated while GPU multiple risk is crowded.' Exit if NVDA orders or hyperscaler capex indicate GPUs are still taking the lion's share of AI budgets.
• Overweight GOOG and META versus a generic AI semi basket if you want the second-order margin beneficiary trade; these names can gain even if hardware vendors mean-revert because lower AI cost structure expands platform economics. Treat this as a 6-18 month structural call rather than a next-quarter catalyst.
• Set an alert on memory pricing and 2026 capex commentary from major memory suppliers; if supply comes back faster than expected, reduce MU exposure aggressively because the downside is a fast multiple de-rating, not a slow grind.