BAE Systems to sell remaining stake in Air Astana

BAE Systems plans to sell its remaining 6.85% stake in Air Astana (terms undisclosed), after previously reducing its holding to 16.95% at the 2024 IPO and selling an additional 10.1% in Dec 2025; Air Astana said the divestment will increase free float. Air Astana reported FY2025 revenue of $1,453.9m (+11.4%) and EBITDAR of $321.2m (+0.8%) with a 22.1% margin and operates 62 aircraft across Air Astana and FlyArystan. The airline trades under ticker AIRA on the Kazakhstan Stock Exchange, Astana International Exchange and London Stock Exchange.

Analysis

Sustained, large-scale OEM demand for high-end AI GPUs elevates the incumbent vendor’s pricing power and forces tighter allocation of constrained wafer and module capacity. That allocation creates a two-tier market: the GPU vendor and integrated appliance buyers capture most incremental margin while third-party server OEMs and smaller AI infrastructure suppliers see mix and ASP pressure over the next 6–18 months.

Second-order winners include software/platform providers that monetize GPU-heavy workloads (higher ARR cadence) and foundries with spare capacity that can pick up overflow at premium pricing; losers are the mid‑tier ODM/EMS players that rely on commodity server ASPs. A key structural risk is policy or foundry disruptions — a single export-control action or a TSMC capacity pivot can reprice forward earnings by 20–40% within a quarter.

Catalysts to watch are enterprise AI capex cadence (reports over the next 2–3 quarters), AMD/Intel product ramps at comparable perf/W (6–12 months), and inventory signals from ODMs (weekly ship-fill and billings). The consensus underestimates optionality both ways: upside if OEMs continue to internalize higher‑margin systems and downside if macro-driven capex pause triggers a fast unwind of channel inventory.

From a portfolio construction standpoint, prefer concentrated, convex exposure to the GPU vendor while explicitly hedging the policy/foundry tail; avoid unhedged long positions in mid‑tier server OEMs that can flip from beneficiary to casualty as procurement shifts from OEMs to integrated suppliers.

Key Decisions for Investors

• Long NVDA via a 6–9 month call spread: buy 6–9m ATM+10% calls and sell 6–9m ATM+35% calls to capture continued pricing power while capping premium outlay. Target asymmetric payoff: 2–3x premium if OEM demand sustains; max loss = premium. Tighten or unwind if implied vol falls >30% or quarterly guidance misses.
• Hedge policy/foundry tail: buy NVDA 12–18 month puts for 25–30% notional protection (or put calendar) sized to cover 30–50% of the NVDA delta exposure. This keeps upside participation while limiting single-event downside risk from export controls or supply interruptions.
• Pair trade (alpha-seeking): long NVDA equity and short a mid‑tier server/ODM (e.g., SMCI-sized exposure) for 3–6 months — size short at 25–40% of the NVDA long. Rationale: captures margin re‑allocation if integrated OEMs internalize systems; stop-loss if ODM reports order acceleration or NVDA loses allocation advantage.
• Selective TSLA exposure: add a directional 9–12 month call (or long equity) sized to 5–10% of tech exposure to play upside from continued access to best-in-class accelerators, but finance with selling near-term calls to fund premium. Set trigger to trim if macro-driven auto demand indicators (monthly deliveries, dealer inventory) weaken materially.