Nokia Stock: Record €2.8B in AI Orders — So Why Did It Still Fall? Why We Say HOLD (NOK Q2 2026)

Nokia (NOK) Q2 2026 — Nokia (NOK, NYSE ADR) reported Q2 & Half-Year 2026 on July 23, 2026. Comparable operating profit rose 18% YoY to €434M — a beat vs the ~€382M LSEG consensus — with comparable operating margin up 70bps to 9.0% and net sales of €4.82B, +8% reported (+9% constant currency). But on a REPORTED basis Nokia posted an operating LOSS of €50M, a −1.0% margin, as it accelerated restructuring (now €800M of charges in 2026); reported diluted EPS was €0.00 vs comparable €0.07. The story was AI & Cloud: order intake hit a record €2.8B and sales to AI/cloud customers grew 105% (more than doubled), driving Network Infrastructure +12% (Optical Networks +20%, boosted by the 2025 Infinera acquisition; IP Networks +16%). But only ~half of that order intake converts to revenue within 12 months, and Nokia guided comparable operating profit 'largely flat' Q2→Q3 before a Q4 jump — management's own 'lumpy' caveat. The larger Mobile Infrastructure segment (55% of sales) grew just 6% with operating profit flat at €310M and margin slipping to 11.6%. Net cash was €2.78B (~$3.2B, only ~6% of market cap); the Board declared a €0.04/share dividend (~1.6% ADR yield). FY26 outlook: comparable operating profit €2.1–2.6B ('somewhat above midpoint'), Network Infrastructure net sales +12–14%, capex €800–900M. The ADR — which quadrupled off a $4 low to a $17.45 high on the AI-networking theme — has round-tripped back to ~$9.97 (−3% on the day, ~43% off its high). At ~18× EV/comparable-EBIT and ~25× comparable earnings, our EV/EBIT through-cycle frame lands fair value near $9.00 — modestly below today's price and well below the Street's BUY consensus and ~$17.5 average target. Our call: HOLD, 3/5 — deliberately more cautious than Wall Street. Nokia spent the last two years as the market's favorite AI-networking comeback: the ADR quadrupled off a $4 low to a $17.45 high as Wall Street decided the old Finnish telecom-gear maker was a clean way to play the AI data-center buildout. Q2 2026, reported July 23, was supposed to be the vindication — and on the surface it was. Comparable operating profit jumped 18% to €434M, beating the ~€382M consensus; net sales grew 8% to €4.82B; and the headline number was electric: a record €2.8 billion of AI & cloud order intake, with sales to those customers more than doubling (+105%). Optical Networks (supercharged by the 2025 Infinera acquisition) grew 20% and IP Networks 16%. And yet the stock fell ~3% on the day and sits ~43% below its high. Why? Because the report has two very different profit numbers: strip away the 'comparable' polish and Nokia posted a REPORTED operating LOSS of €50M — a negative 1.0% margin — dragged down by €800M of accelerated restructuring. Dig further and the growth is narrower and lumpier than the headline: only about half the €2.8B order book converts within a year, management guided profit 'largely flat' into Q3, and the bigger half of the company — Mobile Infrastructure, 55% of sales — grew just 6% with flat profit and slipping margins. So the AI-datacenter engine is real, but it's one smaller segment inflecting while the majority sits still, funded by a thin ~€2.8B net-cash cushion (only ~6% of the market cap) and a small ~1.6% dividend. New CEO Justin Hotard (ex-Intel data-center chief) is doubling down — pushing capex to €800–900M and buying NXP's Arizona chip campus to make Nokia's own optical semiconductors — an Intel-style vertical-integration bet whose payoff is 2027–2028, not now. The honest question isn't whether AI demand is real (it clearly is) but whether it becomes durable, smooth, compounding profit — or stays a series of big, uneven, hard-to-model quarters the market has already handed a growth multiple. At ~18× EV/EBIT and ~25× comparable earnings, an EV/EBIT through-cycle frame lands fair value near $9 — modestly below today's ~$10 and far below the Street's BUY / ~$17.5 target. Our call: HOLD, 3/5 — real AI traction, already in the price, and we're deliberately more cautious than Wall Street. Not financial advice. THE CALL: HOLD (3/5, REAL AI TRACTION, ALREADY IN THE PRICE — A FULL MULTIPLE ON ONE LUMPY GROWTH ENGINE) — base-case value ~$9.00 vs ~$9.97 today. What to watch: proof the AI order surge is durable rather than lumpy — specifically REPORTED operating margin turning clearly positive as the €800M restructuring rolls off, and the €2.8B AI/cloud order book converting into smooth, compounding revenue (not another flat-then-lumpy quarter) — which would flip us to a buy; the risks that would turn us more negative are Mobile Infrastructure staying flat, order conversion slipping, or the capex-heavy Arizona/optical semiconductor bet swallowing the thin net-cash cushion, any of which would pull fair value toward the $8s Also on YouTube: @ChargedAlpha DISCLAIMER: For informational and educational purposes only. Not financial advice. Do your own research before any investment decision.