The 3 Best Artificial Intelligence (AI) Stocks for September

Nvidia’s fiscal Q2 revenue grew 106% YoY and the company guided to ~70% revenue growth for fiscal 2028, which the article frames as a major positive demand signal for AI GPUs. Micron benefits from a memory-chip shortage with elevated pricing, while the article notes production buildouts with new facilities coming online in 2027-2028. Amazon’s AWS capex is cited at up to $220B this year, with AWS growth accelerating from 28% YoY in Q1 to 37% YoY in Q2, and AWS contributing 60% of operating income—supporting a bullish outlook for profits over the next few years.

Analysis

NVDA remains the cleanest read-through on the AI capex cycle, but the second-order beneficiaries may be more interesting than the headline leader: memory, interconnect, and advanced packaging are where incremental scarcity converts into pricing power. The market usually overpays for near-term revenue growth and underprices the risk that hyperscalers eventually optimize around custom silicon and inference efficiency, which can slow GPU growth before overall AI spend rolls over.

MU looks like the better asymmetry because memory is a classic shortage market with operating leverage to spot pricing and low visibility on supply response. The key risk is cyclical overbuild: once 2027-28 capacity comes online, margins can compress quickly, so the trade only works if pricing stays tight longer than consensus expects. Watch for any sign of contract mix weakening or inventory normalization; that would be the first falsifier.

AMZN is a slower-burn winner: the current capex surge is depressing free cash flow today, but it should increase AWS utilization and push more revenue into high-margin infrastructure over the next 1-3 quarters. The market may still be undervaluing how much incremental AWS growth flows through to operating income, but if cloud growth slips back toward the mid-20s or spending remains elevated without measurable utilization gains, the rerating thesis stalls. Relative to NVDA, AMZN has the cleaner multi-quarter setup because the multiple can expand on margin leverage even if top-line growth merely stays strong, not heroic.

Key Decisions for Investors

• Long MU vs. short NVDA for 1-3 months: express the view that memory scarcity is under-owned relative to the crowded AI leader trade. Risk/reward is favorable if memory pricing stays firm; falsify if NVDA re-accelerates guidance or MU gross-margin commentary stops improving.
• Buy AMZN on pullbacks for a 6-12 month horizon: the catalyst is AWS operating leverage, not retail growth. Use a 15-20% upside target and reduce if AWS growth falls below ~30% for two consecutive quarters or capex rises without utilization evidence.
• If already long NVDA, trim into strength and sell 1-2 month covered calls against core exposure. The upside remains real, but the incremental re-rating is less attractive than MU or AMZN unless the next earnings print shows another step-up in forward guidance.