Nokia is not a broken AI trade. It is a misclassified one. The shares have doubled over twelve months — $4.60 on 4 September 2025 to $10.03 on 4 September 2026, a gain of 118% — and are still up 55% year to date. Yet they sit 40.5% below the record close of $16.85 set on 2 June 2026, and closed at $9.77 on 3 September, the lowest since April. The company that produced that chart also raised full-year guidance in July. Both things are true, and the reason they are true at the same time is the single most under-reported number in Nokia’s Q2 2026 accounts.
That number is the operating margin of the division everyone is buying the stock for. Network Infrastructure — the optical and IP business selling directly into AI data-centre interconnect — grew sales 12% to €2,037 million and delivered €166 million of operating profit. That is an 8.1% margin. Mobile Infrastructure, the legacy business the market treats as a melting ice cube, produced €310 million on €2,680 million: an 11.6% margin. In other words, every euro of mix shift towards the AI story dilutes group profitability. Nokia is growing into a lower-margin business, and the market has been re-rating that fact since June. Our $16 bull case and $7 bear case both start from this arithmetic.
Key facts: Nokia (NOK) at a glance
- Share price: $10.03 at the 4 September 2026 close; 52-week range $4.48–$17.45 — stockanalysis.com
- Performance: +118% over twelve months and +55% year to date, but 40.5% below the 2 June 2026 record close of $16.85
- Q2 2026 net sales: €4,815 million, up 8% reported and 9% in constant currency (Q2 and half-year 2026 report)
- Q2 2026 comparable operating profit: €434 million, up 18%, at a 9.0% margin (up 70 basis points); comparable EPS €0.07, up 75%
- Cash: free cash flow of negative €732 million in Q2; net cash fell from €3,788 million to €2,776 million in a single quarter
- AI and cloud: €2.8 billion of order intake in Q2, with sales more than doubling year on year; roughly half of orders convert to revenue within twelve months
- FY2026 guidance (raised): comparable operating profit €2.1–2.6 billion, up from €2.0–2.5 billion; Network Infrastructure sales growth 12–14%; capex €800–900 million; restructuring charges €800 million
- Strategic holder: Nvidia invested $1 billion in October 2025 for the AI-RAN and 6G partnership; the disclosed stake was valued at $2.21 billion in late August 2026
What actually happened in Q2 — and why the stock fell anyway
The quarter itself was good. Net sales of €4,815 million rose 8% as reported and 9% in constant currency. Comparable operating profit of €434 million was up 18% and beat the roughly €382 million consensus, lifting the margin 70 basis points to 9.0%. Comparable earnings per share rose 75% to €0.07. For the half year, comparable operating profit of €735 million was up 28% and EPS of €0.13 up 63%. Management raised the full-year comparable operating profit range to €2.1–2.6 billion and guided Network Infrastructure to 12–14% sales growth. Chief executive Justin Hotard’s summary was that “Q2 demonstrates our strategy is delivering results,” with the company “on track to deliver somewhat above the midpoint.”
Investors sold it anyway, and the reason sits two lines below the profit number. Free cash flow in the quarter was negative €732 million, and net cash fell by €1,012 million, from €3,788 million to €2,776 million. A company whose operating profit is compounding at 18% consumed a billion euros of cash in three months. That is what an equipment order ramp looks like from the inside: components bought and built into inventory long before a customer pays for a deployed network. Roughly half of the €2.8 billion of AI and cloud orders booked in Q2 converts to revenue within twelve months — which means the cash goes out this year and comes back next.
The second problem is the component that is consuming the cash. Memory pricing has become a live input cost across telecom hardware. Ericsson fell nearly 12% on 14 July 2026 after flagging exactly that pressure, dragging the peer group with it, and Nokia’s own de-rating dates from the same window. As Yahoo Finance noted after the results, the market sold the outlook rather than the quarter, with Q3 operating profit guided broadly flat against Q2.
The competitive and customer response
Nokia’s most important relationship is now with a chip company, not a carrier. Nvidia invested $1 billion in October 2025 to build an AI platform for 6G, pairing Nvidia’s AI-RAN work with Nokia’s radio access portfolio; by late August 2026 the disclosed value of that holding had risen to $2.21 billion. The strategic logic is that the radio network becomes an inference workload, and the base station becomes a small data centre. That is a 2028–2030 revenue story, not a 2026 one, and it is the main reason the stock carries a higher multiple than its telecom-equipment history would justify.
The nearer-term demand is in glass and routers. Optical Networks grew 20% and IP Networks 16% in Q2, both well ahead of the 12% divisional growth rate, which by arithmetic means the remaining fixed-access lines grew far more slowly. Hyperscalers buying data-centre interconnect are the customers behind those two lines, and they are the reason Nokia raised its forecast for AI and cloud growth to a 27% compound annual rate through 2028, from 16% previously.
That demand is also being reshaped by policy. A draft FCC ban on Chinese optical transceivers has begun repricing the entire optical supply chain in Western vendors’ favour — a dynamic we set out in detail in our analysis of the draft FCC transceiver rules and in the AAOI bull and bear case. Nokia is one of the few vendors with the scale to absorb the volume that displaces, but displaced volume arrives with the same memory bill everyone else is paying.
Meanwhile the power constraint that governs how fast Nokia’s customers can build is being solved elsewhere in the stack — through on-site generation of the sort covered in our Fermi (FRMI) analysis and the fuel-cell route in our Bloom Energy (BE) prediction. Networking demand is downstream of that: no power, no cluster, no interconnect order.
Market impact: the arithmetic behind $16 and $7
Begin with the mix problem, because it is the analytical core of this piece and almost nobody has run it. Network Infrastructure earned an 8.1% operating margin in Q2 (€166 million on €2,037 million). Mobile Infrastructure earned 11.6% (€310 million on €2,680 million). Guidance calls for Network Infrastructure to grow 12–14% this year while mobile grows mid-single digits. Mechanically, group margin faces a mix headwind at exactly the moment the AI narrative is strongest. The 70 basis points of margin expansion delivered in Q2 came from cost discipline and Nokia Technologies licensing, not from the AI-facing hardware.
The bull case at $16. This is a retest of the 2 June record close, not a new valuation regime, and it needs three things. First, the €2.8 billion quarterly AI and cloud order intake has to persist and convert, taking FY2027 comparable operating profit towards €3 billion against the €2.1–2.6 billion guided for 2026. Second, the working-capital build has to reverse — a single quarter of strongly positive free cash flow would neutralise the bear argument outright, because the cash outflow is the market’s evidence that this growth is uneconomic. Third, memory pricing has to stop rising, or the AI-RAN partnership with Nvidia has to start producing dated, quantified 6G revenue. Get two of the three and $16 is reachable; the stock traded there ninety days ago.
The bear case at $7. That is roughly where Nokia traded in February and March 2026, before the spring re-rating, and it requires no operational collapse. It requires only that memory cost inflation compresses hardware gross margin through 2027, that the mix shift towards 8.1%-margin Network Infrastructure continues, and that free cash flow stays negative for another two or three quarters while €800 million of restructuring charges and €800–900 million of capex are absorbed. In that scenario, the €2.1–2.6 billion operating profit guide holds but the quality of it is questioned, and the multiple returns to where a cash-consuming telecom equipment vendor has always traded.
| Bull case ($16) | Bear case ($7) |
|---|---|
| AI and cloud order intake sustains near €2.8bn a quarter and converts on schedule | Memory cost inflation compresses hardware gross margin through 2027 |
| Working capital unwinds; free cash flow turns firmly positive in H2 2026 | Free cash flow stays negative while €800m restructuring and €800–900m capex are absorbed |
| FY2027 comparable operating profit approaches €3bn against €2.1–2.6bn guided for 2026 | Mix keeps shifting to 8.1%-margin Network Infrastructure from 11.6%-margin mobile |
| Nvidia AI-RAN partnership produces dated, quantified 6G revenue | 6G remains a 2029 story with no revenue attached, and the strategic premium deflates |
The synthesis worth keeping: Nokia’s twelve-month chart and its cash flow statement tell opposite stories, and the market resolved the contradiction in favour of the cash flow statement in June. Anyone underwriting the bull case is really underwriting a working-capital reversal, not a demand forecast. The demand is already in the order book.
Regulation, geopolitics and the licensing floor
Two regulatory forces pull Nokia’s valuation in opposite directions. The first is protective: Western restrictions on Chinese network equipment and, more recently, the draft FCC rules on Chinese optical transceivers, systematically remove low-cost competition from Nokia’s addressable market. That is a durable structural benefit and it is the reason Nokia’s optical growth is outrunning the market’s.
The second is fiscal and political: European operators remain under margin pressure and under regulatory obligation to invest in coverage, which caps the pricing power of their suppliers. Nokia’s Q3 guidance of broadly flat operating profit against Q2 is a statement about carrier budgets as much as about component costs.
Underneath both sits Nokia Technologies, the patent licensing arm, which is the least discussed and most valuable stabiliser in the group. Licensing revenue is high-margin, contractual and largely indifferent to memory prices, and it is a meaningful contributor to the 9.0% group margin achieved in Q2 despite the hardware mix. When the bear case argues for $7, it is implicitly arguing that hardware losses can overwhelm a licensing annuity — which is a higher bar than a simple cyclical de-rating.
What happens next: three predictions
1. Q3 free cash flow is the number that moves the stock, not revenue. Nokia has already guided Q3 operating profit to be broadly flat. The variable that is not guided, and that fell €732 million in Q2, is cash. A positive print re-opens the path to the June highs; a second consecutive large outflow makes the $7 scenario the base case rather than the bear case.
2. Memory cost commentary becomes standard telecom-equipment disclosure by Q4. Ericsson flagged it in July and the whole group de-rated together. Expect vendors to start quantifying memory exposure per unit and hedging it contractually, the way handset makers learned to in previous DRAM cycles. The first vendor to disclose a hedged position will be rewarded for it.
3. The Nvidia partnership gets a revenue number attached during 2027, or the premium goes. A $1 billion investment that has appreciated to $2.21 billion is a validation of the stock price, not of a business line. AI-RAN currently contributes no disclosed revenue. Investors have been patient because the order book elsewhere is strong; that patience is a function of the share price, and it thins every month the stock spends below $10.
Having tracked the optical and telecom supply chain through the 2026 AI build-out, our position is that Nokia is the cheapest genuine exposure to AI data-centre interconnect in the listed market — and that “cheap” is doing real work in that sentence, because the reason it is cheap is a mix and cash problem the bulls have not answered. Readers weighing the same trade at the compute layer will find the counterpart in our Nvidia bull and bear analysis.
Frequently asked questions
What is the Nokia NOK stock prediction for 2027?
Our framework sets a $16 bull case and a $7 bear case against the $10.03 close of 4 September 2026. The bull case is a retest of the 2 June record close and needs FY2027 comparable operating profit heading towards €3 billion with free cash flow turning positive. The bear case needs no collapse — only continued memory cost inflation, continued negative free cash flow, and the margin mix shift towards Network Infrastructure that is already underway.
Why has Nokia stock fallen 40% since June 2026?
The de-rating began in the week Ericsson fell almost 12% on memory cost inflation, on 14 July 2026, and continued through the Q2 report. Nokia beat on profit — €434 million comparable operating profit, up 18% — but reported negative free cash flow of €732 million and guided Q3 operating profit broadly flat. The market sold the outlook and the cash statement rather than the quarter.
How much of Nokia’s business is exposed to AI data centres?
Network Infrastructure, which houses the optical and IP businesses selling into data-centre interconnect, generated €2,037 million of Q2 sales, up 12%, with optical up 20% and IP up 16%. Nokia booked €2.8 billion of AI and cloud orders in the quarter, with roughly half expected to convert to revenue within twelve months, and has raised its AI and cloud growth forecast to a 27% compound annual rate through 2028.
What did Nvidia’s investment in Nokia buy?
Nvidia invested $1 billion in October 2025 to co-develop an AI platform for 6G, combining its AI-RAN technology with Nokia’s radio access portfolio. The disclosed stake was valued at $2.21 billion in late August 2026. The partnership currently contributes no separately disclosed revenue; it is a 2028–2030 story that supports today’s multiple.
Is Nokia profitable and does it pay a dividend?
Yes. Q2 2026 comparable operating profit was €434 million at a 9.0% margin, with comparable EPS of €0.07, and the company guides to €2.1–2.6 billion of comparable operating profit for the full year. It also carries net cash of €2,776 million, though that fell by more than €1 billion during Q2 as working capital absorbed the order ramp.
Which Nokia division has the better margin?
Mobile Infrastructure, at 11.6% in Q2 2026 (€310 million on €2,680 million), out-earns Network Infrastructure at 8.1% (€166 million on €2,037 million) — despite the latter being the AI-exposed growth engine. Because guidance calls for Network Infrastructure to grow 12–14% while mobile grows more slowly, group margin faces a structural mix headwind even if every division performs as planned.
This article is analysis, not investment advice. Prices and scenario levels are as of the 4 September 2026 close.