NVIDIA Reportedly Ends GeForce RTX 5070 Ti Production, RTX 5060 Ti 16 GB Next

ASUS reports that NVIDIA has effectively ended production of the GeForce RTX 5070 Ti due to severe supply constraints, driving retail scarcity and price increases, and is now phasing out the RTX 5060 Ti 16 GB in favor of an 8 GB variant as GDDR7/DRAM shortages bite. The planned RTX 50 “SUPER” refresh (higher-memory SKUs) was postponed and is not currently scheduled, prompting AIB partner frustration and signaling near-term product roadmap disruption that could compress mid-range SKU availability and alter pricing dynamics across partners and retailers.

Analysis

Market structure: The forced EOL of NVDA's RTX 5070 Ti and shrinking 5060 Ti 16GB supply shifts pricing power to remaining NVDA SKUs and benefits AMD (uses GDDR6) and PC OEMs that can source alternative memory. Expect mid-range ASPs to rise 10–25% in the next 30–90 days as channel scarcity manifests; GDDR7 spot tightness will bid up DRAM suppliers’ revenues by mid-2026 if unresolved. Cross-asset: DRAM suppliers (MU) and equipment names (LRCX) get positive commodity tailwinds, NVDA equity may see elevated IV and tactical weakness, while USD/NTD flows could tighten vs. USD if Taiwan AIBs repatriate cash to buy memory.

Risk assessment: Tail risks include a prolonged GDDR7 shock (>6 months) that forces SKU rationalization, AMD responding with aggressive pricing, or export controls tightening China access—each could swap market share by >5ppt within 12 months. Immediate (days) risk is volatile retail pricing; short-term (3–6 months) depends on fabs’ memory ramp; long-term (12+ months) risk is structural product mix shift toward lower-memory SKUs. Hidden dependency: AIB channel inventory opacity can mask true sell-through; catalyst to reverse is a Micron/Mass production ramp or public AIB order disclosures.

Trade implications: Tactical: trim NVDA exposure 3–5% and establish a 2–4% long in AMD (AMD) and 1–2% speculative long in INTC targeting share gains in low-end GPUs over 3–12 months. Pair: long AMD vs short NVDA (equal notional) for 3–6 months. Options: buy NVDA 3-month 10% OTM put spreads (paid cost <2% notional) to hedge near-term downside; consider long MU 6–12 month call spread to play DRAM tightness.

Contrarian angles: Markets may over-penalize NVDA ignoring that data-center GPU demand (>60% revenue) cushions mid-range SKU losses and could sustain margins; a 10% NVDA pullback could be overdone if DC bookings stay healthy. Watch for unintended consequence: shorter-memory SKUs could push gamers to AMD/INTC, accelerating software ecosystem shifts—if AMD captures +2–4ppt GPU share in 4 quarters, re-rate opportunities emerge.

Key Decisions for Investors

• Reduce NVDA (NVDA) net long exposure by 3–5% of portfolio within 1–2 weeks; hedge remaining NVDA exposure with a 3-month 10% OTM put spread (pay no more than ~2% notional premium) to protect vs. an 8–12% downside move.
• Establish a 2–3% long position in AMD (AMD) over 1–3 weeks to capture potential mid-range share gains and price-sensitive demand; add if AMD outperforms NVDA by >5% relative in 30 days.
• Open a pair trade: long AMD / short NVDA equal-dollar for a 3–6 month horizon, target relative return >8% and exit if NVDA recovers within 10% of pre-trade price or catalyst (Micron ramp) confirmed.
• Buy MU 6–12 month call spread (bullish on DRAM tightness) sized 1–2% of portfolio; exit or trim if GDDR7 spot prices decline >15% QoQ or Micron announces meaningful capacity additions.
• If willing to be tactical, sell covered calls on NVDA against existing holdings (1–2 month tenors) at 5–10% OTM to monetize elevated IV until memory supply signals normalize (monitor Micron/Mem supplier commentary within 30–90 days).