R&S Group H1 2026 slides: record backlog amid sales moderation

R&S Group reported H1 2026 net sales of CHF 179.2m (-13.1% YoY), driven by FX headwinds (CHF 4.4m), shifted deliveries (CHF 9.6m), and divestiture impacts, but maintained a 19.0% EBITDA margin (CHF 34.0m). The company increased its order backlog 17.1% to a record CHF 357.9m and guided to FY2026 sales of CHF 410-420m with an EBITDA margin of 19-21% despite near-term utility normalization. Free cash flow rose 121% to CHF 11.5m (6.4% of sales), and the dividend policy remains CHF 0.50/share; shares declined modestly in the article context on the backdrop of delayed utility orders.

Analysis

This is more a credibility hit to AMD’s AI-share narrative than a meaningful earnings event for either name. A high-profile customer picking NVDA reinforces that in accelerators, the moat is increasingly software ecosystem, deployment certainty, and time-to-rack, not just raw chip specs; that tends to widen the valuation gap when investors are already paying for future share gains.

The first-order revenue impact to NVDA is likely small unless the procurement expands into a multi-quarter cluster build, but the second-order winners are the surrounding NVDA stack names that benefit from standardized deployments and follow-on networking/server attach. AMD’s risk is multiple compression, not an immediate EPS reset; the market may start discounting slower-than-modeled share capture across 2026 AI capex budgets.

Contrarian view: the move may be overdone if investors are extrapolating one customer decision into broad market-share conclusions. Without order size, this is an attention-grabbing reference account, not proof of a durable shift; if AMD prints any credible large design win or ROCm adoption improvement into the next earnings cycle, the selloff should retrace quickly. Falsifiers for the bearish AMD read are a disclosed hyperscaler/defense win or a materially better AI revenue guide; for NVDA, supply constraints or a broader capex pause would cap upside.

Key Decisions for Investors

• Tactically short AMD / long NVDA for 1-3 months as a relative-value pair, sized modestly; thesis is multiple divergence rather than fundamental revenue surprise. Cover if AMD announces a large AI design win or reaccelerating data-center guidance.
• If entering after the open, prefer to sell AMD strength rather than chase the gap lower; the headline is strong sentiment fuel but weak cash-flow evidence. Stop if AMD reclaims the pre-event trading range on volume within 3-5 sessions.
• For investors who want cleaner exposure, favor NVDA over AMD in semiconductor baskets for the next earnings cycle; the signal supports NVDA’s ecosystem premium and keeps AMD’s AI ramp under a higher proof threshold.
• Avoid aggressive options positioning on NVDA here unless order size is confirmed; the direct incremental revenue is likely too small for a high-conviction call buy. A call spread financed by a small AMD put spread is a cleaner expression if implied vol is bid.