Nvidia stock rating reiterated at Buy by BofA ahead of GTC event

BofA reiterated a Buy on NVDA with a $300 price target ahead of GTC and highlighted product roadmap items (Feynman GPUs, co-designed/disaggregated products, proprietary optics); Tigress raised its NVDA target to $360 and kept a Strong Buy. Nvidia trades at a P/E of 36.68 and PEG of 0.54, with BofA forecasting a $0.5 trillion cumulative Blackwell ramp and consensus data-center revenue of ~$0.75 trillion (2026-27) rising to ~$1.0 trillion (2027-28). CEO Jensen Huang announced a $30 billion investment in OpenAI; Broadcom received a reaffirmed Buy from Aletheia and Bank of America flagged Marvell amid optics tailwinds, while VCI launched Malaysia’s first NVIDIA-powered AI GPU computing center.

Analysis

The current narrative around AI leadership is amplifying value upstream in optics, switch ASICs, and server OEMs; companies with proprietary photonics IP and scale manufacturing capacity will likely capture disproportionate incremental margin as hyperscalers prioritize latency and rack-level integration. Expect 6–18 month supply-side bottlenecks (advanced packaging, HBM, co-packaged optics) to convert product announcements into staggered revenue recognition rather than an instant demand surge, creating discrete re-rating windows when procurement/qualification milestones are hit.

Market pricing is effectively a binary frontier: flawless execution preserves a high multiple, while any 1–2 quarter cadence slip or model-efficiency improvement (smaller inference compute per query) can produce outsized multiple compression — a 5–10% miss in data-center growth could translate to 20–30% downside in investor-implied market cap in the short term. Large strategic minority investments into single private AI platforms concentrate liquidity and governance risk; an IPO delay or valuation reset of that private asset would feed back into hardware multiple volatility.

Given the asymmetric exposure across the stack, the highest-expected-value opportunities are not pure hardware momentum longs but staged, catalyst-aware positions: owning modular exposure to networking/optics winners and selected OEMs while hedging headline-driven delta in the hyperscaler-facing GPU maker. Vol trades around major conferences (GTC-like events) are attractive — buy-side volatility compresses quickly post-announcement, favoring short-dated premium sellers with defined hedges.

Key Decisions for Investors

• Go long MRVL (Marvell) 6–12 months: buy stock on strength or buy a 9–12 month call spread. Thesis: capture optics/SerDes share without paying halo multiple. Position sizing: 2–4% portfolio; target 30–40% upside; stop-loss 18–20%.
• Construct a paired trade: Long AVGO (Broadcom) vs short NVDA to reduce headline beta. Notional: 1.0 AVGO per 0.6–0.8 NVDA (adjust to beta), timeframe 3–9 months. Reward: AVGO’s diversified cash returns vs NVDA execution sensitivity; risk: NVDA continues to re-rate higher — cap exposure to 3% portfolio.
• Event-driven NVDA volatility play: sell 30–45 day covered-call or iron-condor after the conference once headlines are digested. Collect IV with wing buys to define downside (e.g., 15–20% OTM protective put). Expected IRR: collect IV >2–4% monthly; tail risk defined by bought wings.
• Staged exposure to server OEMs (SMCI or VCIG for regionals): buy on 8–15% pullbacks tied to supply-chain confirmations, hold 6–12 months. Rationale: OEMs win follow-on server builds; target 25–50% upside on execution; stop 20%.
• If you want pure asymmetric long on the GPU narrative, use long-dated LEAP calls (12–24 months) sized 1–2% portfolio and hedge with short-dated puts sold against realized volatility after each major disclosure. This keeps upside participation while monetizing near-term IV; be prepared to roll hedges if execution misses occur.