SpaceX, Tesla to continue ordering Nvidia chips at scale- Musk By Investing.com

Tesla reiterated development of its in‑house AI5 chip (optimized for Optimus and robotaxi) and said Terafab will launch March 21; Tesla signed a $16.5B deal with Samsung for chip builds but reports indicate volume AI5 production may be delayed to mid‑2027. Musk confirmed SpaceX and xAI continue to rely on Nvidia (H and Blackwell lines) and SpaceX plans a public offering this year and aims to build AI data centers in space — positive for Nvidia demand but with company‑specific execution and timing risks.

Analysis

NVIDIA’s position in the AI stack creates a two- to three-year cascade of demand across foundries, advanced packaging (2.5D/3D), HBM suppliers and datacenter integrators; any incremental bespoke silicon from OEMs is more likely to re-shape where value accrues in the supply chain than to eliminate hyperscaler spend. Expect pockets of tightness (EUV node slots, HBM lanes, high-density interposers) to persist into 2026-2028, which will amplify pricing power for incumbents that control end-to-end supply relationships.

Execution risk is concentrated and fast-moving: a missed tape-out, a foundry yield problem, or fresh export controls can truncate growth expectations in weeks; conversely, hyperscaler refresh cycles or a surprise enterprise capex rebound can accelerate revenue recognition inside a single quarter. Watch near-term cadence triggers (product architecture announcements, foundry guidance, and large hyperscaler bookings) on a 0–90 day window and structural capacity indicators (tool orders, HBM lead times) over 6–18 months.

Consensus underestimates a common pattern in semiconductor history: vertical OEM chips typically coexist with platform incumbents and often expand TAM via new endpoints rather than replace core workloads — this is a likely mid-term outcome here. The main contrarian risk is valuation fragility: current multiples price in near-perfect scaling across both cloud and on-device robotics; a single quarter of decelerating cloud AI capex would compress realized upside by 30–50% for high-multiple names, so position sizing and optionality are essential.

Key Decisions for Investors

• Long NVDA directional via a 9–12 month call spread (buy near-the-money call, sell ~30–35% OTM call) entered within the next 2 weeks to capture near-term product/capacity catalysts; max loss = premium paid, target gross return 40–80% if datacenter demand sustains, breakeven if growth reaccelerates beyond current consensus.
• Pair trade to isolate execution risk: go long NVDA and short TSLA on a delta-adjusted, equal notional basis for 3–9 months. Rationale: hedge out broader AI enthusiasm and capture divergence between infrastructure demand and consumer/autonomy execution; set a 20% stop on the short leg and take profit on the pair if NVDA outperforms TSLA by >30% (histor outperformance window).
• Macro tail hedge: buy a 12–18 month SPX 5%–10% OTM put or a VIX call spread sized to cover 3–5% portfolio drawdown risk over 6–12 months. Rationale: protects against the UBS-style extreme geopolitical/market shock that would compress high-multiple tech irrespective of semiconductor-specific fundamentals.
• Reallocate exposure within the supply chain: tactically trim high-beta system integrators/OSATs on strength and redeploy into NVDA exposure via capped-cost option structures; if HBM lead times shorten or foundry yields improve (6–12 months), rotate back into selected integrators for catch-up upside.