Nokia

Phase A — Understand the business

Lens 1 · Company Overview

Nokia Corporation (Espoo, Finland; NYSE ADS: NOK; ~5.742bn shares outstanding at 31 Dec 2025 ) sells the physical and software infrastructure that carries the world's communications and, increasingly, AI-datacenter traffic. It earns money four ways (FY2025 reportable segments):

• Network Infrastructure (NI) — EUR 7,986m net sales, EUR 780m operating profit (9.8% margin). The growth engine and the AI story. Three units: Optical Networks EUR 3,019m (+85% YoY, EUR 1,273m of that from the Infinera acquisition), IP Networks EUR 2,594m (flat), Fixed Networks EUR 2,373m (+3%). Sells coherent optical transport, IP routing, and fiber/fixed-wireless to telecom operators, hyperscalers ("AI & Cloud") and "Mission Critical Enterprise & Defense."
• Mobile Networks (MN) — EUR 7,806m net sales, EUR 220m operating profit (2.8% margin). Radio access (RAN) across 2G–5G, the legacy core and the perennial margin problem. Down 4% YoY; margins compressed by a EUR 120m one-time contract settlement (2019-vintage project).
• Cloud and Network Services (CNS) — EUR 2,606m net sales, EUR 338m operating profit (13.0% margin). Core networks, enterprise campus, automation software. The quiet over-performer — operating profit +64% YoY on favorable mix.
• Nokia Technologies (patents) — EUR 1,501m net sales, EUR 1,059m operating profit (70.6% margin). The cash cow: a licensing business monetizing Nokia's cellular/multimedia IP estate. ~EUR 1.4bn contracted run-rate. Near-100% gross margin, but lumpy (2024 had >EUR 400m catch-up sales, so 2025 fell 22%).
• Group Common and Other — EUR 17m sales, EUR (373)m operating loss — corporate center.

Contract structure / payment terms: mostly project- and order-backlog-driven (order backlog EUR 19.5bn at YE2025 ) plus recurring patent royalties (Technologies). Not take-or-pay; revenue is tied to telecom and now hyperscaler capex cycles. AI & Cloud generated EUR 2.4bn of orders in 2025 — small relative to ~EUR 20bn group revenue but the fastest-growing demand pool.

The structural change to know: at the November 2025 Capital Markets Day Nokia announced a new strategy and collapsed the four segments into two — Network Infrastructure and Mobile Infrastructure — effective 1 January 2026. FY2025 is therefore the last year reported under the old four-segment model. Several "attractive but non-core" businesses were moved into a "Portfolio Businesses" bucket for strategic alternatives, and defense activities were carved into a dedicated incubation unit.

Lens 2 · Supply Chain

Upstream inputs → Nokia → end customer, with named stakeholders:

• Upstream silicon (the chokepoint). Nokia's datacenter switching and optical DSPs ride on third-party merchant silicon. Its new 7220 IXR H6 switch is built on Broadcom's Tomahawk 6 chipset. Broadcom "controls over 80% of high-end switching silicon and supplies the PAM4 DSPs in virtually every 800G/1.6T transceiver". Single-source dependency on Broadcom for the leading-edge switching/DSP layer — Nokia is an integrator on top of Broadcom, not a silicon owner (contrast: it does own optical components via the Infinera acquisition, and is "investing in new fab capacity for our optical components" ).
• Coherent optics / DSP. Post-Infinera, Nokia has in-house coherent DSP and optical engines (Infinera's ICE6/ICE7 line) and ships 800G ZR/ZR+ coherent pluggables. This is the vertical-integration claim management leans on ("vertical integration across hardware and software").
• NVIDIA (new strategic node, both supplier and investor). Nokia is porting 5G/6G RAN software onto NVIDIA's Arc Aerial RAN Computer platform. NVIDIA becomes both a technology partner and a 2.90% shareholder.
• Manufacturing. Contract/outsourced manufacturing plus owned facilities (e.g., the new Oulu, Finland 5G/6G campus, 3,000 staff ).
• Downstream customers. Telecom operators globally (AT&T, T-Mobile US, Deutsche Telekom, etc.), hyperscalers / "AI & Cloud" (named obliquely; Meta is a known 800G buyer split across Cisco/Ciena/Nokia ), and Mission Critical Enterprise & Defense (utilities, transport, NATO-adjacent).

Chokepoints: (1) Broadcom merchant-silicon dependency at the switching layer; (2) telecom-operator capex concentration (Mobile Networks still ~39% of revenue and demand-cyclical); (3) FX — an approximately 4–5% negative FX impact on net sales across NI/MN/CNS in 2025 because costs are EUR-heavy and sales USD-heavy.

Lens 3 · Competitive Advantages (moats)

Where the moat is real:

• Coherent optical transport — a duopoly. Post-Infinera, "Ciena operates in a duopoly with Nokia-Infinera in the transport layer". Coherent DSP, line systems and the long-haul/DCI optical layer have high switching costs, multi-year qualification cycles, and only a handful of credible vendors. This is Nokia's best asset and the heart of the AI-infra thesis.
• Patent estate (Technologies). ~EUR 1.4bn/yr of high-margin licensing from foundational cellular/multimedia IP. A genuine, durable, ~70%-operating-margin annuity — though it requires continuous litigation to enforce (the Amazon dispute, resolved March 2025, is the template ).
• Bell Labs / R&D depth. EUR 4,855m R&D in 2025 (24.4% of sales); 100-year-old Bell Labs. Real, but expensive — R&D intensity this high is also a symptom of a sub-scale competitor that must spend to stay in the game.

Where the moat is thin:

• Mobile / RAN. This is a brutal 3-way oligopoly (Ericsson, Nokia, Huawei) where Nokia is the perennial #2/#3 in the West, lost the marquee AT&T deal to Ericsson in 2023, and runs 2.8% segment margins. No pricing power; demand is operator-capex-gated. The NVIDIA AI-RAN partnership is an attempt to manufacture a moat here (be the AI-native RAN leader) — promising but unproven.
• Datacenter switching. As a Broadcom-silicon integrator, Nokia competes with Arista, Cisco, and white-box ODMs on a layer where it has no silicon advantage. It is a challenger, not an incumbent.

Bargaining power: weak-to-moderate over customers (telecom operators are concentrated, sophisticated buyers; hyperscalers are even more powerful and multi-source deliberately). Weak over its key supplier Broadcom. Strongest leverage sits in the patent business, where Nokia is the licensor.

Lens 4 · Segments

FY2025 vs FY2024, all ``:

Geography (Network Infrastructure, the growth segment): Americas EUR 3,688m (+35%), APAC EUR 1,648m (+16%), EMEA EUR 2,650m (+12%). The Americas surge — driven by Infinera + AI&Cloud + Mission-Critical — is the most important geographic fact: North America is where the AI re-rating is being earned, and where Nokia "increased our market share" per the CEO. Mobile Networks, by contrast, fell in the Americas (−9%) and APAC (−5%, Greater China weakness).

Trend & cause: the mix is shifting toward the high-multiple stuff (optical/AI) and away from low-margin radio — but slowly. In FY2025, Optical+IP+Fixed (NI) = ~40% of revenue; Mobile = ~39%; the high-margin patents annuity = ~7.5%. The accelerating line (Optical +85%, even ex-Infinera +organic) is the bull case; the EUR 1,059m patents operating profit (70.6% margin) is what actually carries group profitability — Technologies alone is larger than the entire group's EUR 885m reported operating profit, meaning the operating businesses ex-patents barely break even. That is the single most important number in the segment table.

---

Phase B — Measure performance

Lens 5 · Earnings Result

Latest annual (FY2025, the 20-F):

• Net sales EUR 19,889m, +3% (vs EUR 19,220m); ~4–5% FX headwind, so mid-single-digit organic.
• Reported operating profit EUR 885m (4.4% margin), down hard from EUR 1,970m (10.2%) — but this is distorted: 2024 included EUR 1,514m of Technologies operating profit with catch-up licensing; 2025's EUR 1,059m is the cleaner run-rate. Management guides on "comparable" operating profit ~EUR 2.0bn, in line with guidance.
• Profit for the year EUR 660m; basic EPS EUR 0.12.
• Free cash flow EUR 1,465m (down from EUR 2,021m); operating cash flow EUR 2,071m.
• Balance sheet: cash & equivalents EUR 5,462m; net cash EUR 3,378m (gearing −16%) — a fortress, even after EUR 1.7bn of acquisitions, EUR 759m dividends and EUR 624m buybacks. Order backlog EUR 19.5bn.
• R&D EUR 4,855m (24.4% of sales) — rising in absolute and % terms.

Latest quarterly print — Q1 2026, reported 23 April 2026:

• Net sales ~EUR 4.5bn, +4% comparable cc.
• Comparable operating profit EUR 281m, +54% YoY — beat consensus (~EUR 250m). Margin +200bps.
• AI & Cloud net sales +49%; EUR 1bn of new AI&Cloud orders (Optical-led).
• Guidance RAISED: Network Infrastructure FY2026 net-sales growth lifted to 12–14% (from 6–8%); Optical+IP combined to 18–20% (from 10–12%). FY2026 comparable operating-profit guide EUR 2.0–2.5bn.
• Stock surged on the print.

Read: the operating businesses are inflecting (NI accelerating, CNS profitable, MN stabilizing), the patents line is lumpy-but-huge, and the guidance raise is the cleanest bull signal in the file. The flag: reported group profitability is thin (4.4% margin) and depends on the patent annuity; strip Technologies and the hardware/software businesses run low-single-digit margins.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on disk (transcripts/ empty), so ``. Tone trajectory across the last ~6 quarters:

• 2023 (Lundmark era): defensive, cost-cutting — "losing the AT&T deal hurts," gross-cost-out program of EUR 800m–1,200m launched, 5G market "challenging."
• 2024: stabilization, balance-sheet repair, Infinera deal announced — pivot language begins ("infrastructure underpinning the AI supercycle").
• 2025 (Hotard arrives April): decisive re-framing. Recurring new phrases: "AI supercycle," "AI & Cloud," "connecting intelligence," "AI-native networks," "increase our organizational clock speed." The CEO's signature line — "Nokia changed the world once by connecting people. Now we can change it again by connecting intelligence" — is a deliberate narrative reset from telecom to AI-infrastructure.
• Q1 2026: confident, guidance-raising, AI&Cloud front-and-center.
• What they stopped saying: the apologetic 5G-headwinds framing of 2023. The story is now offense, not defense.

Caution: this is exactly the sentiment arc you'd expect from management trying to engineer a multiple re-rating. Tone has shifted faster than the revenue mix. Treat the AI framing as partly real (orders, Optical growth) and partly narrative management.

Lens 7 · Comps

Peer table. Multiples are ``; where not found, marked n/a. Do not treat the trailing P/E spread as precise — aggregators disagree by a wide margin on Nokia's trailing P/E (26 to 103 depending on whether discontinued ops / one-offs are included).

Read: every credible optical/networking peer trades on a rich multiple right now (Ciena ~173x, Cisco fwd ~27x, Nokia fwd ~37x) because the whole group has been bid up on the AI-infrastructure trade. Nokia at fwd ~37x / EV/EBITDA ~28x is not cheap on any absolute measure and is priced above Cisco despite materially lower margins and ROE (Nokia's ROE has never exceeded ~6% in three years vs Cisco's high-teens-plus). The bull retort: Nokia's forward EPS is depressed (turnaround base) so the multiple overstates richness; on 2028 targets the multiple compresses. The bear retort: you're paying a premium-growth multiple for a 3% group operating margin and a 3% ROE.

Lens 8 · Stock-Price Catalysts (>5% moves, last ~5 years)

`` throughout:

• Dec 2023 — AT&T contract loss to Ericsson. Nokia plunged to a 3-year low (Helsinki −7% on the day); cut its 2026 comparable operating-margin target to ≥13% from ≥14%. AT&T was a single-digit % of sales but ~$1.6bn of lost MN revenue — the defining negative catalyst of the era and the trigger for the cost-out program.
• 2024 — AT&T resolution + fiber deal. EUR 150m Q2'24 net-sales/OP benefit from the contract resolution; September 2024 a new AT&T fiber deal. Stabilization.
• June 2024 — Infinera acquisition announced (closed Feb 2025). Re-rating begins on the optical/AI-DCI thesis.
• Feb 2025 — Justin Hotard named CEO (ex-Intel DC&AI / ex-HPE HPC). Market read it as an AI-pivot signal.
• Oct 2025 — NVIDIA $1bn investment / AI-RAN partnership. Major positive catalyst; the explicit "AI infrastructure" stamp from the AI bellwether.
• Nov 2025 — Capital Markets Day: new strategy, two-segment reorg, 2028 targets (EUR 2.7–3.2bn comparable OP). Re-rating consolidated.
• 2025–2026 — analyst chase. Stock +109% in 12 months; JPMorgan PT to $21 from $14 (Overweight). ATH $16.85 on 2 June 2026.

Pattern: historically Nokia reacted to operator contract wins/losses and margin-target changes (a telecom-capex stock). Since late 2024 the reaction function has changed — the stock now moves on AI-infrastructure validation events (Infinera, NVIDIA, hyperscaler orders). That reaction-function shift IS the re-rating, and it is the thing that can reverse fastest if AI-order momentum stalls.

---

Phase C — Judge people & books

Lens 9 · Management

• CEO: Justin Hotard (since 1 April 2025). Track record: 25 years in tech; HPE 2015–2024, rising to EVP/GM High-Performance Computing, AI & Labs — delivered the world's first exascale supercomputer (Frontier) for the US DoE, a genuine flagship credential for the AI-infrastructure pivot. Then Intel EVP/GM Data Center & AI Group — a mixed scorecard: shipped Xeon 6 (Granite Rapids/Sierra Forest), but Gaudi-3 AI accelerators missed sales targets (much of which predated him). Net: a credible AI/datacenter operator parachuted in precisely to execute the AI-infra repositioning — the appointment is the strategy. Tenure is short (~14 months), so the track record at Nokia is still being written.
• Predecessor: Pekka Lundmark (2020–2025) — ran the painful 5G stabilization, the cost-out program and the Infinera deal; handed Hotard a cleaned-up balance sheet and the optical platform.
• Skin in the game / insider ownership: no insider-transactions.csv on disk → n/a. As a widely-held Finnish large-cap with no founder/controlling shareholder, insider ownership is low; NVIDIA's new 2.90% strategic stake is the most notable holder signal.
• Capital allocation: disciplined and shareholder-friendly. FY2025: EUR 759m dividends + EUR 624m buybacks + EUR 1.73bn M&A (Infinera) + EUR 501m to buy out the Nokia Shanghai Bell minority — all funded from cash while keeping EUR 3.4bn net cash. Proposed dividend EUR 0.14/share; active buyback continues. ROE/ROIC are the weakness, not the willingness — ROE 3.1% (FY25), 6.2% (FY24), 3.2% (FY23); ROCE 4.3%. Capital is returned and reinvested competently, but the underlying business doesn't yet earn a high return on it.
• Red flags: none egregious. The Infinera deal is an all-share-heavy, dilutive acquisition (127.4m new ADSs issued ) plus the NVIDIA directed issuance (166.4m new shares at $6.01 ) — two large equity issuances in one year mean real dilution; share count is climbing. Not a governance red flag, but a per-share-value watch item.
• Archetype: professional managers, not founders — appropriate for a turnaround-via-execution story. The bet is on operational competence and strategic clarity, not founder vision.

Lens 10 · Forensic Red Flags

Grounded in the 20-F financial statements and notes ``:

• Reported vs "comparable" operating profit gap. Reported OP EUR 885m vs the ~EUR 2.0bn "comparable" figure management guides on — a ~EUR 1.1bn bridge of restructuring, PPA amortization (Infinera) and one-offs. The headline guidance metric is a heavily-adjusted non-IFRS number. Legitimate for a restructuring company, but watch that "comparable" doesn't permanently exclude recurring restructuring (the cost-out program runs through 2026 with restructuring charges ≈ annual savings).
• Profit quality concentrated in one segment. Nokia Technologies (patents) earns EUR 1,059m operating profit — more than the entire group's reported EUR 885m. The operating/hardware businesses ex-patents are ~breakeven-to-thin. A weak licensing year (as 2025 was, −22%) swings the whole P&L. Licensing revenue is lumpy and litigation-dependent (>EUR 400m of 2024's Technologies sales were catch-up; the Amazon settlement closed a multi-year dispute).
• One-time settlement noise in Mobile Networks. A EUR 120m net-negative contract settlement (2019-vintage project) hit 2025 MN gross profit; 2024 had a +EUR 150m AT&T accelerated-revenue benefit. Underlying MN margin is even thinner than the 2.8% headline once you normalize these.
• Cash flow vs earnings: healthy — operating cash flow EUR 2,071m comfortably exceeds reported net income; FCF EUR 1,465m. No divergence red flag. Working capital released cash. Receivables increased only EUR 25m; inventories fell EUR 149m. No receivables/inventory blow-out — clean.
• Receivables factoring (disclosed): "Nokia sells trade receivables and customer loan receivables to various financial institutions primarily without recourse in the normal course of business". Standard for the industry but it flatters reported DSO/working capital — a routine watch item, not a red flag.
• Provisions: litigation & environmental provision EUR 289m (of which environmental EUR 130m). Warranty provision additions driven by the 2019-project contract settlement. Restructuring provision EUR 371m.
• Deferred tax / unused losses: EUR 21,918m of unused tax losses/credits with no deferred tax asset recognized (mostly France) — a legacy of years of losses; means low cash tax in some jurisdictions but signals how much value was historically destroyed.
• SBC / dilution: the real per-share risk is share issuance (Infinera 127.4m + NVIDIA 166.4m new shares) rather than ordinary SBC. Flagged in Lens 9.

Regulatory findings (required sub-section).

• SEC Litigation Releases: none. AAERs: none. Verified via EDGAR EFTS (LR + AAER), period 2021-06-21 → 2026-06-21.
• Non-SEC enforcement (web search, "Nokia" + FTC/DOJ/FDA/consent decree/settlement/fine/penalty): no current material new enforcement action surfaced in the search window. Historical context (not in the EDGAR scope): Nokia/Alcatel-Lucent's legacy FCPA matters were resolved years ago; nothing new material found for 2025–2026. Treat as clean pending a deeper search.
• 10-K-equivalent (20-F Note 6.1 "Legal matters") — the company's own disclosure, quoted: "While management does not expect any of the legal proceedings it is currently aware of to have a material adverse effect on Nokia's financial position, litigation is inherently unpredictable..." Specific matters: Brazil mass labor litigation (former employees post-managed-services exit; majority settled); ~250 US asbestos-related matters (legacy premises/products/contractor liability); IP litigation — the Amazon patent dispute was resolved by agreement in March 2025.
• Conclusion: No material regulatory or accounting-enforcement findings — verified via SEC EDGAR EFTS (LR, AAER), web search, and 20-F Note 6.1 as of 2026-06-21. The forensic flags are quality-of-earnings issues (adjusted-metric reliance, single-segment profit concentration, dilution), not integrity issues.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection

Build from FY2025 actuals + the company's own FY2026 guidance + 2028 CMD targets. All outputs ``; inputs labeled.

Anchors:

• FY2025 comparable operating profit ~EUR 2.0bn; reported EPS EUR 0.12.
• FY2026 guide: comparable OP EUR 2.0–2.5bn; NI net sales +12–14%; Optical+IP +18–20%.
• 2028 target: comparable OP EUR 2.7–3.2bn; FCF conversion 65–75%; NI net-sales CAGR 6–8% 2025–2028; Group Common opex cut to EUR 150m (from ~EUR 350m).
• Share count rising (~5.74bn + NVIDIA's 166m issuance + Infinera dilution) → use ~5.9bn diluted for forward years.

Base / Bull / Bear comparable-EPS path (comparable OP → tax ~22% effective [implied from FY25 profit-before-tax EUR 915m → profit EUR 660m, ~28% but distorted; use 22% normalized] → minorities small → diluted by share count). EUR EPS, then ADR ~1:1.

• FY2026 (base): comparable OP ~EUR 2.25bn (guide midpoint) → after-tax ~EUR 1.75bn → ÷ ~5.9bn sh → comparable EPS ~EUR 0.30. Bull ~EUR 0.33 (top of guide, share-count held); bear ~EUR 0.27 (low guide + dilution).
• FY2027 (base): OP ~EUR 2.6bn (mid-path to 2028) → EPS ~EUR 0.34. Bull EUR 0.40 (AI&Cloud orders compound, NI 14%+); bear EUR 0.28 (telecom capex stalls, MN drags).
• FY2028 (base): OP ~EUR 2.95bn (CMD midpoint) → after-tax ~EUR 2.3bn ÷ ~5.9bn → EPS ~EUR 0.39. Bull ~EUR 0.46 (top CMD target EUR 3.2bn + buyback-shrunk share count); bear ~EUR 0.30 (AI theme under-delivers, group OP ~EUR 2.3bn, dilution).

Valuation cross-check: at ~$13.49 ADR (~EUR 12.4 at ~1.09 USD/EUR ), the base FY2028 EPS ~EUR 0.39 implies a 2028 P/E ~32x even after three years of target delivery — i.e., the current price already discounts successful execution of the 2028 plan. To justify the price on a 15x "AI-infra" terminal multiple you need FY2028 comparable EPS closer to EUR 0.80, roughly double the company's own 2028 target. The math says the stock prices the bull case as the base case.

(Per the --watchlist rule, no forecast.ts create is run in this loop. If promoted to a thesis, log the base FY2028 comparable-OP forecast — "NOK FY2028 comparable operating profit ≥ EUR 2.95bn," p≈0.45 — as the scoreable call.)

Lens 12 · Bull vs Bear

Bull case. Nokia is the cheapest credible way to own the optical and AI-RAN layers of the AI build-out. The optical-transport duopoly with Ciena is a structurally advantaged position; Infinera gives Nokia owned coherent DSP/optical engines and North-American hyperscaler relationships; AI&Cloud orders compounded to EUR 2.4bn in 2025 and grew 49% in Q1'26; the NVIDIA partnership + investment is a once-in-a-cycle validation that puts Nokia inside the 6G/AI-RAN standard. Management raised NI guidance to 12–14% and has a credible operator (Hotard, the Frontier-exascale builder) executing a clean two-segment simplification. The patents annuity (EUR 1.4bn, 70% margins) funds the whole transition. If AI&Cloud goes from 8% to 20%+ of revenue and the group re-rates from a telecom multiple to even a discounted infrastructure multiple, the equity compounds. JPMorgan's $21 PT embodies this.

Bear case (permanent-impairment risks). (1) It's still a low-return telecom company. Group operating margin 4.4%, comparable ~10%, ROE 3%; ~92% of revenue is telecom radio, IP, fixed and patents — slow, cyclical, capex-gated. The AI part is ~8%. (2) The re-rating already happened. +109% in 12 months, fwd P/E ~37, EV/EBITDA ~28 — priced above higher-margin Cisco. The math (Lens 11) shows the price discounts full 2028-target delivery; there is little margin of safety and a real risk of a multiple reset back toward telecom-equipment norms (~12–15x) if AI-order momentum stalls for even two quarters. (3) Silicon dependency + hyperscaler power. Nokia integrates Broadcom silicon and sells to multi-sourcing hyperscalers who will commoditize the box; the durable value may accrue to Broadcom/NVIDIA, not Nokia. Pre-mortem (18 months out, thesis broke): AI&Cloud orders plateau as hyperscalers standardize on white-box + Broadcom/Marvell silicon and Ciena takes the marquee 800G slots; Mobile Networks stays sub-scale and capex-starved; a weak licensing year hammers reported profit; the stock de-rates 40–50% from a ~37x forward multiple back toward Ericsson-like levels even as revenue holds — a multiple collapse, not an earnings collapse. Contrarian view of what the market refuses to see: that this is a competent, cash-rich, shareholder-friendly value-grade industrial that the market has temporarily repriced as a growth-grade AI play — and the gap closes by the multiple falling, not the business failing.

Lens 13 · Devil's Advocate (short-seller)

• What structurally breaks the money machine? A multiple de-rating. You don't need the business to fail — you need the market to remember it's a 4% operating-margin telecom supplier. At fwd ~37x the asymmetry is negative.
• Revenue concentration / shift: Mobile Networks (~39% of sales) is hostage to operator capex and lost the flagship AT&T radio deal to Ericsson once already. The "growth" is concentrated in Optical/AI&Cloud (~8% of sales) — if that demand pool normalizes or shifts to merchant-silicon white boxes, the entire growth narrative evaporates while the legacy base keeps shrinking.
• Why the moat is weaker than bulls think: in datacenter switching Nokia is a Broadcom-silicon integrator with no silicon edge against Arista/Cisco/ODMs. In RAN it's the perennial #2/#3. Only the optical-transport duopoly and the patent estate are genuinely defensible — and Ciena (with a $7bn backlog ) is gunning for the same 800G hyperscaler sockets.
• Most dangerous competitor bulls underestimate: Broadcom (and Marvell) — the silicon owners who could disintermediate the box vendor — and Ciena for the optical sockets. Also Arista in DC switching.
• Worst capital-allocation / accounting concerns: two big equity issuances in one year (Infinera + NVIDIA) dilute per-share value; reliance on a heavily-adjusted "comparable operating profit" that bridges ~EUR 1.1bn over reported; EUR 21.9bn of unrecognized tax losses memorializing historical value destruction.
• What must hold for today's price: AI&Cloud orders keep compounding ~40%+, NI hits 12–14% growth and then 6–8% CAGR to 2028, the patent annuity holds ~EUR 1.4bn, and the market keeps assigning a ~35x+ forward multiple for three more years.
• If growth disappoints 20–30%: NI growth halves to ~6–7%, AI&Cloud orders flatten, comparable OP lands ~EUR 2.3bn in 2028 (low end), and the multiple compresses to telecom norms → a plausible 40–55% drawdown from current levels, mostly via de-rating.
• Single permanent-impairment scenario: hyperscalers standardize AI networking on merchant-silicon white boxes (Broadcom/Marvell + ODMs), structurally bypassing branded box vendors. Plausibility: moderate over 3–5 years for switching; lower for coherent optical transport (where qualification and DSP IP still gate entry). This is the scenario that would permanently re-rate Nokia back to a value-grade industrial.

Lens 14 · Management Questions (ordered by information value)

1. AI&Cloud is ~8% of revenue but ~all of the growth narrative. At what revenue mix does AI&Cloud become large enough that the group margin and growth profile structurally change — and what's the realistic year?
2. In datacenter switching you're integrating Broadcom's Tomahawk 6 silicon. Where exactly is Nokia's durable value-add versus a white-box ODM running the same silicon — and why won't hyperscalers commoditize the box?
3. Walk through the bridge from FY2025 reported operating profit (EUR 885m) to "comparable" (~EUR 2.0bn). Which of those adjustments are genuinely one-time versus recurring through the 2026 cost program — and what does reported margin look like in 2028?
4. The NVIDIA partnership: beyond the $1bn equity, what are the binding commercial commitments? What AI-RAN revenue is contracted versus aspirational, and what's the T-Mobile trial gating to a deployment decision?
5. Nokia Technologies earns more operating profit than the entire group. How do you de-risk the licensing P&L from its lumpiness and litigation-dependence — and what's the renewal/run-rate visibility through 2028?
6. Optical Networks grew 85% (mostly Infinera). What is the organic Optical growth and the organic AI&Cloud order trajectory, stripping the acquisition?
7. Ciena has a ~$7bn backlog aimed at the same 800G hyperscaler sockets. Where are you winning versus Ciena at named hyperscalers, and what's your design-win share at 800G/1.6T?
8. Two large equity issuances in 2025 (Infinera + NVIDIA). What is the share-count trajectory through 2028, and how does the buyback offset dilution on a per-share basis?
9. Mobile Networks runs 2.8% margins and is sub-scale versus Ericsson/Huawei. Under the new two-segment "Mobile Infrastructure," what is the credible path to acceptable returns — and is exit/partnership on the table?
10. ROE has been ~3–6% for three years. What ROIC does the 2028 plan actually generate, and when does Nokia earn its cost of capital?
11. Which "Portfolio Businesses" are you exploring strategic alternatives for, on what timeline, and what proceeds/margin-accretion do they imply?
12. You're investing in owned optical-component fab capacity. What's the capex envelope, the make-vs-buy logic versus merchant DSP, and the payback?
13. FX was a 4–5% net-sales headwind in 2025 (EUR cost base, USD sales). How structurally hedged are you, and how much of "growth" is FX-masked?
14. The defense activities were moved into a dedicated incubation unit. What is the scope, the ambition, and the revenue/margin profile you're targeting there?
15. If the AI-infrastructure demand pool normalizes in 2027, what is the downside revenue/margin scenario you're managing against — and what's the floor for the dividend and buyback?

---