LSL FY25 profit in line, sees further growth in 2026

LSL reported FY2025 underlying operating profit of £32.6m (vs. £32.2m consensus), ROCE of 35% and net cash of £27.8m; the board declared a total dividend of 11.4p and announced a £12m share buyback. Management maintained FY2026 outlook and said trading is in line so far; Surveying revenue rose 8%, Financial Services was flat, and the Pivotal JV contributed £1.7m (vs. £1.0m estimate). Pivotal Growth generated revenue approaching £100m, completed 24 acquisitions to date, repaid shareholder loans and needs no further cash; LSL is pursuing a pipeline targeting 50 lettings-book acquisitions over three years.

Analysis

Sustained, high-volume GPU demand from vertically integrated customers materially reorders the AI hardware value chain: fabs and advanced packaging (TSMC, Amkor/ASE) become the choke points for growth, not system integrators. That rebalancing elevates capex elasticity—small shifts in fab capacity or export policy can swing available GPU supply by 10-30% within a quarter, creating outsized volatility for the GPU vendor and any customers who lack multi-sourcing options.

For large buyers of accelerators, continued external GPU purchases are a double-edged sword. On one hand, buying off-the-shelf accelerators accelerates product cycles and feature delivery; on the other, it leaves them exposed to vendor allocation and ASP moves, forcing either higher OPEX for capacity or heavy capex to internalize compute (which takes 12–36 months to be meaningful). That timing mismatch creates a multi-year runway where GPU vendors capture disproportionate economic rents before buyers can realistically substitute away.

Consensus appears to underweight two second-order outcomes: (1) the margin/composition squeeze at GPU vendors if they broaden into lower-ASP segments to satiate demand, and (2) the growing bargaining power of hyperscale/cloud customers who will pay up front for guaranteed allocation. Both can compress forward multiples if capex cycles normalize or if a major buyer signals a strategic pivot to custom silicon. Near-term catalysts to watch are fab allocation updates, export-control headlines, and any public timelines from large buyers on internal ASIC rollouts.

Key Decisions for Investors

• Long NVDA via 9–18 month call calendar or LEAP call + short nearer-dated OTM calls (ticker: NVDA). Rationale: capture upside from sustained accelerator demand while using call-write to finance premium; target asymmetric 2:1 upside potential vs defined premium loss (max loss = premium paid). Add more on pullbacks into 5–15% IV spikes tied to supply headlines.
• Pair trade — long NVDA / short TSLA (over 6–12 months). Rationale: NVDA benefits immediately from strong order flow while Tesla faces optionality risk (in-house ASICs/Dojo) that could remove a multi-year GPU demand stream. Position size to target 1.5–2x sensitivity to NVDA moves vs TSLA to create positive carry if NVDA outperforms by >20%; set stop-loss if NVDA falls >25% or if Tesla issues concrete in-house rollout timing.
• Protective hedge for platform exposure: buy TSLA 6–12 month put spread (small width) funded by selling short-dated calls (ticker: TSLA). Rationale: inexpensive insurance against a fast pivot to internal silicon or a demand slowdown; limits downside while keeping some upside participation.
• Event-monitor: set alerts for TSMC capacity comments, US export-control updates, and supplier inventory/margins (Micron, Amkor). If any signal shows a >20% change in effective GPU supply assumptions, re-rate exposure within 48–72 hours.