Sidetrade : Croissance à deux chiffres de la rentabilité sur H1 2026

Sidetrade reported H1 2026 revenue of €34.8M, up 19% reported and 21% at constant currency, while EBIT rose 22% to €5.7M and net income increased 15% to €4.7M. New annual recurring revenue bookings more than doubled to a record €5.16M (+112%), already 20% above all of 2025, with total contract value up 194% to €18.71M. AI-native products generated €1.02M of Q2 ARR, or 33% of quarterly bookings, supporting management's expectation that the commercial rebound will begin contributing materially to revenue from Q4 2026 and accelerate in 2027.

Analysis

ALBFR’s investable inflection is not the reported revenue growth but the conversion of materially stronger bookings into recognized subscription revenue from Q4 onward and through 2027. The higher new-logo mix and longer initial commitments should improve net retention visibility and reduce dependence on cross-sell, while North American bookings create a geographic mix shift toward the market most likely to support a higher enterprise-software multiple. The key question for the next two quarters is whether this translates into billings/deferred-revenue growth rather than merely elevated TCV, which is sensitive to contract duration.

The margin narrative requires adjustment before assigning a full AI-software premium. Consolidated gross margin has fallen, and the headline operating result benefits materially from the research tax credit; the much stronger like-for-like margin excludes the acquired business that investors must nevertheless underwrite. Proprietary infrastructure may protect inference economics versus token-priced SaaS peers if agent usage scales, but owned GPU capacity also introduces utilization risk: underused H100/H200 assets turn a claimed cost advantage into fixed-cost deleveraging. The company’s competitive edge is most credible in regulated, data-sensitive receivables workflows, less so in broadly customizable enterprise-agent deployments where hyperscalers and ERP vendors have distribution advantages.

Near term, the stock can re-rate on the October Q3 release if subscription growth, backlog conversion and North America momentum confirm the order intake. Over 6-18 months, success depends on proving that AI-native products raise ARPU and retention without cannibalizing legacy modules; production deployment of beta products in Q4 is the first real commercialization test. Consensus may overvalue the AI branding and undervalue the balance-sheet/earnings-quality constraints: net debt remains meaningful for a small-cap issuer, and a lower tax credit, integration drag, or a weak Q3 revenue bridge would compress a premium multiple quickly in a relatively illiquid Euronext Growth name.

Key Decisions for Investors

• Initiate a small, liquidity-aware long ALBFR only after the October 20 Q3 release confirms accelerating reported subscription growth and disclosed deferred-revenue/RPO conversion; target a 6-12 month re-rating on 2027 revenue visibility. Do not add solely on ARR/TCV headlines; reduce if Q3 fails to show a clear Q4 revenue bridge or if organic subscription growth remains near mid-single digits.
• Use a 3-6 month long ALBFR / short ENX sector-proxy overlay only for investors seeking idiosyncratic exposure with reduced European growth-beta risk. The thesis is operational execution and AI monetization, not a broad exchange/market-volume call; size conservatively given ALBFR liquidity and the imperfect hedge.
• Set a Q4 monitoring trigger for AI product production launches: add only if management discloses paid deployments, incremental ARPU or usage economics and stable subscription gross margin. Evidence that agent contracts are bundled into existing spend, rather than additive revenue, falsifies the margin-expansion case.
• Avoid treating NVDA as a meaningful read-through trade. ALBFR’s GPU deployment is too small to affect NVDA fundamentals; the relevant NVDA sensitivity is indirect—higher GPU cost or capacity requirements could pressure ALBFR free cash flow and challenge its fixed-price AI pricing model over 2027.