Nokia Q2 profit climbs 18% as AI data-centre demand doubles cloud sales

Nokia posted Q2 2026 comparable operating profit of €434m, up 18% YoY and above LSEG’s €382m analyst forecast. Net sales rose 8% to €4.82bn (9% at constant currency), supported by customers ramping AI data centre builds. The profitability and revenue beat together suggest a favorable near-term demand backdrop for Nokia’s infrastructure exposure.

Analysis

This reads less like a single-company beat and more like another data point that AI capex is broadening from GPUs into the transport layer. The incremental money is likely flowing to vendors with exposure to optical, routing, and high-capacity interconnect, which should matter more for margin mix than top-line growth alone; in other words, the second-order winner is whoever sells picks-and-shovels for data-center connectivity, not the hyperscalers themselves.

The market risk is that investors extrapolate one quarter of better execution into a durable demand inflection before the order book proves it. If this is simply a timing shift in customer procurement, the upside in networking names could fade in 1-2 quarters as inventory normalizes; the tell will be backlog, book-to-bill, and whether managements across the group raise full-year guide rather than just cite AI demand. A weaker macro tape would also expose any carrier-spend dependence still embedded in the model.

Contrarian take: the consensus may be underestimating how much of the AI buildout is still concentrated in a few hyperscalers, which means supplier-level revenue can be lumpy even when end-demand is strong. The right lens is relative quality: names with pricing power, recurring upgrade cycles, and cleaner exposure to data-center interconnect should outperform legacy telecom vendors that merely catch the same thematic bid. If the next print does not show continued order conversion, this move should be treated as a tactical rerate, not a new secular leg.

Key Decisions for Investors

• Long CIEN vs short NOK on a 1-3 month horizon: express the view that AI-driven networking demand accrues more cleanly to a focused optical supplier than to a broader telecom vendor; risk/reward improves if CIEN can show backlog conversion while NOK remains more mixed.
• Add ANET on any post-earnings pullback over the next 2-4 weeks: the market is likely to reward AI infrastructure beneficiaries with stronger operating leverage than incumbent telecom exposure; falsify if hyperscaler capex commentary softens.
• Use a basket trade: long CIEN/ANET, short a legacy telecom capex proxy such as VZ or T if the goal is to isolate AI interconnect spending vs. slower carrier budgets; this is a 3-6 month relative-value expression, not a macro bet.
• If NOK gaps up sharply on the print, consider selling upside via covered calls rather than chasing common stock; the risk is that the market has already priced in a one-quarter AI narrative without evidence of sustained backlog acceleration.
• Watch the next round of hyperscaler capex and 800G/1.6T deployment commentary as the falsifier; if those indicators do not improve, trim exposure to networking beta and rotate into companies with recurring software revenue instead.