Ericsson's New CEO Buys Time: A Great Quarter That Borrows From 2027

Per Narvinger's first call delivered a 7% organic beat and a 47.7% gross margin — and a candid warning that the AI memory crunch turns next year into a transition period.

ERIC-B.ST · Earnings Call · 2026-10-15

The Quarter That Borrowed From Next Year

Per Narvinger opened his first earnings call as Ericsson's CEO with a beat, not a caveat. Organic sales rose 7%, “a strong third quarter where we are executing well in a challenging environment” — Per Narvinger, President and CEO · 2026-10-15, gross margin printed 47.7%, EBITA margin 17%, and every segment grew organically. Network sales jumped 9% year-over-year to SEK 38.7 billion. On the surface, the new boss inherited a machine. Scratch at the quality of that revenue, though, and the story sharpens. CFO Lars Sandstrom laid out three growth building blocks in Networks, and the second one should make you pause: a near-term shift in customer behavior with some customers accelerating spend before price increase. Customers are pulling orders forward because Ericsson has told them the price is going up. Asked point-blank how large that pull-in was, Sandstrom did not quantify it and offered only “It's not super big.” — Lars Sandstrom, Chief Financial Officer · 2026-10-15 A 9% Networks quarter partly built on forward-borrowed demand is a different animal than a 9% quarter built on structural share gains — and the company knows it, because it guided Q4 back to ordinary seasonality.

The Squeeze Between Two Contract Books

The real pressure Ericsson is describing is a scissors: component cost increases ripping upward at the same time it sits on long-term fixed frame agreements on both sides of the business. This is not new — it dominated the prior two calls — but the framing is new. Prior quarters were about mitigation. This quarter is about reclassification. Narvinger split the response into tactical actions — upselling, product substitution, steering the sales mix toward parts of the portfolio with less memory content — and structural action — portfolio optimization, price-book increases, supply-chain moves. His point is that tactical works exactly where it is used, and can only be used so long:

We, as a supplier, cannot absorb these costs, they have to be passed on. But of course, there can be a transition period, and that's what we are saying that '27 could be such a transition period.
Per Narvinger, President and CEO · 2026-10-15

The line that told the market everything came in the final follow-up, when an analyst baited them on the un-mitigated gross-margin hit. Narvinger didn't take it, but he didn't dodge either: “It will have a dilution effect significant if we are not taking action.” — Per Narvinger, President and CEO · 2026-10-15 That is a CEO pre-loading 2027 as a dilution year — for a first call, an unusually honest setup. This is where Ericsson meets the wider tape. The memory squeeze is a global story, not a telecom one: high bandwidth memory and AI data centers names are among the market's 30-day laggards as the AI build-out drains component supply. Ericsson is on the consuming end of that boom, and it is saying so out loud — inventory built to absorb higher component prices and finished goods ahead of Q4. The global keyword pulse and the company's own cash-flow commentary are pointing at the same thing: capital is flowing into AI compute, and squeezing everyone adjacent to it.

What Narvinger Quietly Changed

For a company whose CEO transition was choreographed, the structural moves announced with the print are the freshest company-unique item. Ericsson is dissolving its Enterprise Wireless Solutions business area, folding neutral-host and enterprise 5G back into Networks to concentrate R&D muscle, leaving the former Cradlepoint as a stand-alone unit, and keeping Vonage-descended assets under a mandate to reach black numbers faster. It is consolidating market areas into a single Asia Pacific region and trimming the executive team. The contrast with the prior regime is worth flagging. On the last several calls, Börje Ekholm described Enterprise as a business that cannot be consistently loss-making and would need time — but stopped short of a timeline. Narvinger marks his tenure by moving the pieces, not promising a date. The vocabulary changed too: leaner executive team, R&D synergies, and share gain all entered the top tier of the quarter's language. This is a new CEO re-cutting an organization he inherited.

Optionality: AI RAN and the Sky

The forward-looking half of the call had genuine new substance. Narvinger put numbers on the AI-in-the-network case that his predecessor kept abstract: an AI-native link-adaptation feature delivering a 10% spectral-efficiency gain, deployed commercially with three operators and running on more than 60,000 base stations. He then did the math live — if SpaceX paid roughly $8 billion for spectrum, a 10% efficiency lift is worth about $800 million, and some of that should flow back to Ericsson. “Reaping gains from AI is not about what platform you execute on. It's actually to get the functionality out there.” — Per Narvinger, President and CEO · 2026-10-15 This is a subtle but important pivot. Six months ago, management was telling the market the opposite of a data-center story — “We're not going to see any sales directly from data center expansions right now” — Borje Ekholm, President and CEO · 2026-04-17 — and conceding they were not the frontrunner on the AI wave. Now the claim is narrower but harder: AI value arrives at the edge, over the air interface, where Ericsson's silicon already sits. Whether that converts to price in a flat RAN market is the open question Narvinger inherited and declined to answer with a number. The bottom line. Ericsson just printed its best-feeling quarter in a while, and its new CEO spent the call walking the market back from extrapolating it. Strong sales, a tax-return of pull-in demand, a cost curve driven by a boom the company doesn't own, and a reorg that signals impatience with prior dotting of i's. The tension is not whether the quarter was good — it was. It is whether a supplier that can't absorb component costs can pass them through before the transition period eats a chunk of 2027. Narvinger is betting the answer is yes, and he wants you to grade him on that, not on the beat.