Earnings Highlights: Nokia (NOK) | Q2 FY2026 — AI & Cloud Revenue Doubles, Orders Hit €2.8B as Nokia Recasts Itself as an Optical and Routing Supplier for AI Infrastructure

2 days ago

8 min read

Nokia's most important signal this quarter is not "9% revenue growth" — it is that its new AI & Cloud segment doubled quarterly revenue to €446 million and orders jumped to €2.8 billion. Yet management reined in the excitement itself, stating plainly that more than half of these orders won't convert to revenue for 12 months and shouldn't be treated as the norm. The real story has three threads: Network Infrastructure (optical and routing) is being repriced by AI data center demand; Mobile is placing a long-term bet on the AI-RAN platform unveiled just last week; and supply-chain bottlenecks (memory, indium phosphide) are shifting from a "lead-time problem" into a structural race where whoever secures materials gets to ship.

1. Three Takeaways to Remember First

First, the growth engine has changed hands. What drove the company this quarter was not traditional telecom equipment but AI & Cloud. The segment grew revenue 105% YoY with €2.8 billion in orders, and both Optical and IP are surging, weighted toward Optical. Nokia used to be a company that "follows telco capex"; now it has a second identity that "follows hyperscaler capex."

Second, management is very restrained about orders. CEO Justin Hotard repeatedly stressed two things: orders will be "lumpy," so don't expect this kind of intake every quarter; and orders are "lengthening" — only about half of the €2.8 billion will convert to revenue within the next 12 months, with the rest falling into later periods. That shift in order structure is itself evidence that AI demand is pushing visibility further out. Broadcom said the same thing last quarter, which we broke down in Earnings highlights: Broadcom (AVGO) | Q2 FY26 — $30B of AI orders in one quarter pushes optical visibility out to 2028.

Third, gross margin doesn't yet reflect the growth. AI & Cloud is ramping fast but hasn't translated into meaningful gross margin or operating leverage. Management admitted it directly: as they said at Capital Markets Day, they are investing up front to make the company leaner and more scalable, and the benefits "will show up later." Translation: 1H26 results are still absorbing costs.

2. Revenue and Financials

All figures below are on a comparable basis, with growth rates in constant currency. This quarter Nokia also announced a reporting change: Fixed Wireless Access (FWA) and Enterprise Campus Edge were classified as discontinued operations, and historical figures for 2025 and Q1 2026 were restated. The change reduced Q2 comparable revenue by €66 million but actually added €13 million to comparable operating profit — in other words, what was cut was low-margin business that dragged on profitability.

Overall performance:

Net sales: +9% YoY, driven mainly by Network Infrastructure.
Gross profit: €2.2B, gross margin 46% (+70bps), expansion coming from NI — especially Optical Networks — as Infinera integration synergies continue to kick in.
Operating profit: €434M, operating margin 9% (+70bps). Some software revenue originally expected in Q3 was recognized early in Q2; meanwhile, stock-based compensation (SBC) rose YoY on a higher share price and expanded plans, a 150bps headwind to operating margin.
Free cash flow: -€732M. Q2 is seasonally the weakest cash quarter (annual employee bonuses are paid in Q2), and working capital rose for inventory build.
Net cash position: €2.8B; the balance sheet remains strong.

By segment, Network Infrastructure was the star this quarter, with net sales up 12% YoY:

Optical Networks +20%: dual engines of AI & Cloud customer demand and telco investment in transport networks.
IP Networks +16%: order momentum that began in 2H25 is now converting into revenue.
Fixed Networks -2%: Optical Line Terminal (OLT) +18%, but ONT -16% (a deliberate exit from the low-price segment).
NI gross margin +240bps to 42.7%; operating margin +170bps to 8.1%.

Mobile Infrastructure net sales grew 7% YoY (core software +1%, radio networks +7%, technology standards +15%), with gross margin at 49.3% (boosted by the early software revenue recognition) and operating profit flat YoY. Management guided Q3 MI gross margin back down to 44%-46% (less software contribution), recovering seasonally in Q4.

By customer type, the story is even clearer: AI & Cloud +105%, Telecom +4%, technology licensing +15%. Growth is entirely on the AI side; the telecom market is expected to be "broadly unchanged."

3. Technology and Business Highlights: Orders, Multi-Rail Design, DSPs and AI-RAN

Optical and routing: demand is coming from both scale-out and DCI. Justin spent considerable time separating two things that are easily conflated: the back-end scale-out fabric that connects and expands compute clusters, and data center interconnect (DCI), which runs long-distance, high-bandwidth links between data centers. Both have a routing (IP) component, and routing demand for DCI is growing "very significantly." That is why Nokia is confident saying IP and Optical are growing together. The logic of CPO's three-stage evolution — scale-out moves first, scale-up is the endgame — is covered in more depth in The great optical packaging transition (Part 2): CPO's three-stage evolution — why OBO died, scale-out moves first, and scale-up is the endgame.

First multi-rail ILA design win. This quarter Nokia won its first multi-rail ILA (in-line amplifier) design, using a new optical networking product launched at OFC just this March. Ciena is the most direct comparison for this "system vendors move into AI back-end networks" route — it won the industry's first multi-rail order with Hyper-Rail last quarter; for context see Earnings highlights: Ciena (CIEN) | Q2 FY26 — Hyper-Rail wins the industry's first multi-rail order as system vendors formally enter AI back-end networks.

The four-DSP bet. One analyst questioned whether Nokia can deliver four DSPs by end-2027, a sharp step-up from its past cadence. Justin's explanation: before the merger, Nokia and Infinera were each developing two DSPs, so four was already the combined total; he deliberately "did not cut the DSP teams" because four distinct DSPs can cover different optical-layer needs — from scale-across fabric and DCI to metro and long-haul — giving fuller market coverage than the traditional two-DSP lineup. Timeline: customer trials in 2027, commercial availability by end-2027.

AI-RAN: shifting radio networks from hardware-defined to software-defined. Last week Nokia unveiled what it calls the industry's first commercial AI-RAN platform. The core pitch is "gaining performance through software and AI rather than hardware upgrades" — claiming more than 100% spectral efficiency improvement by 2028 (effectively doubling capacity on existing spectrum) and a software upgrade path to 6G without further hardware spend. The platform is O-RAN compatible and programmable, and Nokia is working with NVIDIA to put GPUs into AirScale. Timeline: pilots by end-2026 (10 public customers, with T-Mobile as the lead US pilot partner), commercial in 2027, volume in 2028. This long-term play of tying compute and optical interconnect into the NVIDIA ecosystem points in the same direction as NVIDIA's recent push to bring the optical supply chain into its own territory. Further reading: NVIDIA buys into optics: when optical communications turn from growth stocks into index heavyweights (2026 W26).

4. Management Outlook: Guidance Unchanged, Q4 Is the Main Event

Nokia maintained its comparable operating profit guidance, with the only adjustment being the technical change of moving FWA and Enterprise Campus Edge into discontinued operations. Management reiterated that full-year operating profit will land "somewhat above the midpoint" of the range.

The quarterly cadence is what to watch:

Q3 net sales: +3%-7% QoQ.
Q3 operating profit: roughly in line with Q2 — because Mobile's software revenue was pulled forward from Q3 into Q2, Q3 loses that piece.
Q4: a marked improvement, driven first by telecom's seasonally heaviest delivery quarter and second by YoY AI & Cloud contribution.

Richard Kramer (Arete) pointed out on the call that if the full year is above the midpoint and Q3 is flat, Q4 profit must be more than double Q2. CFO Marco Wirén's answer: that is simply the normal result of telecom seasonality layered with AI & Cloud, and he stressed that the cash position is comfortable.

Two more numbers to put in the model: free cash flow conversion is now expected at the low end of the 55%-75% assumption (due to added restructuring costs and working capital for inventory build); full-year 2026 restructuring charges of about €800M (covering the wind-down of the 2023-2026 program, which delivered €1.2B in cumulative cost savings, consolidation of China operations, and a new efficiency program mainly affecting Europe).

5. Supply Chain and Customer Clues: Materials Are the Real Variable

Memory is the hardest bottleneck. Asked to rank supply-chain risks, Justin named memory outright — overwhelming demand, shortages, and significant price increases driven by the shortage are a shared pain across the entire tech ecosystem. Nokia's response: lock in supply, simplify designs, reduce specs where possible, and pass costs through to customers.

Indium phosphide (InP): Nokia chooses to build its own. This is where Nokia's long-term ambition shows most clearly this quarter. Its new InP wafer fab in San Jose is already running test wafers, with volume production by year-end; advanced test and packaging capacity at its Pennsylvania site will expand 10x; and today it announced the acquisition of an Arizona fab from NXP to expand InP capacity (though that site won't come online until 2029 at the earliest). InP is the core material of the "engine of light" and an industry-wide recognized bottleneck — Lumentum spoke even more bluntly about the tightness of this same supply chain; see Earnings highlights: Lumentum | JPM 54th Global Technology Conference fireside chat — from boom-bust to structural multi-year growth, InP is the real bottleneck.

Could orders be inflated by "double ordering"? Bernstein's Ulrich Rathe raised the double-ordering risk common in supply-constrained markets. Justin's rebuttal is worth noting: these customers (hyperscalers) are highly sophisticated, and the orders ultimately trace back to advanced optical component manufacturing capacity that customers can audit in person and on which Nokia shares progress transparently — in such an auditable environment, the incentive to double order is low. Nokia has also started asking customers for longer-term commitments.

Customer concentration remains high. On the IP side, the customer base is currently quite concentrated — Justin said "that's how this business is built," going deep with a few large customers first and then gradually expanding the footprint. This is the risk investors need to watch next: growth is strong, but it currently rests on a handful of hyperscale customers.

Co-innovation signals: this quarter Nokia expanded partnerships with Google Cloud (Gemini AI agents in its autonomous networks portfolio), Vodafone Albania (AI network slicing) and Indonesia's Indosat Ooredoo Hutchison (5G modernization plus an AI-RAN upgrade path), and entered an out-of-band management trial inside the data center with a US hyperscaler.

6. Conclusion

This quarter Nokia told a fairly complete transformation story: moving from telecom equipment vendor toward "optical and routing supplier for AI infrastructure + long-term AI-RAN player." The numbers (+9% revenue, 46% gross margin, 9% operating margin) were solid but not spectacular; the real weight lies in the order structure and supply-chain positioning — especially the move to build its own InP fab, which signals Nokia's conviction that control over materials will decide the winners in the coming years.

Three metrics to track next quarter (and through 2H):

1. AI & Cloud revenue conversion pace: how the "half within 12 months" of the €2.8B in orders actually lands, and whether the newer IP line can catch up with Optical.

2. Software-driven swings in Mobile gross margin: whether the Q3 dip to 44%-46% is purely timing as management says, and whether Q4 recovers seasonally.

3. Whether supply can keep up with demand: Justin has said outright that "with more materials, we could probably deliver more revenue" — so the real ceiling isn't demand, it's the supply of InP and memory.

This article is for technology and industry trend analysis only and does not constitute investment advice.