Orange Raises Again — but the Questions Have Moved On

A second EBITDAaL upgrade in two quarters, a new four-operator satellite spectrum consortium, and a conspicuous silence about data centers

ORA.PA · Earnings Call · 2026-10-27

A second raise, in a different room

Orange's Q3 2026 was, on the numbers, the most straightforward quarter a telecom CFO could ask for. Revenue rose 3.6% to EUR 12.2 billion, EBITDAaL grew 4.6%, and operating cash flow climbed 10.3%. Africa and Middle East delivered its fourteenth consecutive double-digit revenue quarter (+14%), Europe 6 accelerated to 5.3%, and France — the perennial worry — grew 0.5% while posting record convergent net adds since 2018 with mobile churn down more than a point and broadband churn down more than two. That was enough for a second upgrade: full-year EBITDAaL growth now expected above 4.5%, having been lifted to "above 3%" back in April. “Q3 confirms the group's strong momentum with revenue growth at 3.6%. EBITDAaL increasing by 4.6% and a double-digit growth of operating cash flow at 10.3%.” — Christel Heydemann, CEO · 2026-10-27 What is more revealing than the numbers is the composition of the question list. Across the 2026 Q1 and Q2 calls the analysts were effectively a merger-arbitrage desk: French competition authority timing, pre-notification, the SFR share-versus-asset structure, MASORANGE net adds, the Vodafone FiberCo perimeter. This quarter those threads are thinner and the new ones are exogenous — agentic AI, the French election, rising rates and the NPV of the SFR synergies, a network outage, and satellite spectrum. The company's story has not changed; the market's anxiety about it has.

The one genuinely new strategic step: a spectrum club

Last week Orange signed a letter of intent with Deutsche Telekom, Telefonica and Vodafone to prepare a potential joint bid for mobile satellite service spectrum being auctioned at the end of 2027 for deployment in 2029. Management frames it as sovereignty infrastructure — a complement, not a competitor, to terrestrial networks — with the spectrum offered wholesale to all European operators. “the bid for the spectrum would then be open and wholesale to all telecom operators in Europe. And we see MSS spectrum really as a complement to our terrestrial network.” — Christel Heydemann, CEO · 2026-10-27 This is a real departure from the posture of prior calls. In April, satellite was a sequence of non-exclusive bilateral tests — Skylo in France, a Starlink trial in Spain, AST in Romania — with the CEO explicit that “we're not contemplating signing a simple MVNO transaction because we really think value comes from the customer experience and the direct-to-customer management” — Christel Heydemann, CEO · 2026-07-28. Moving from opportunistic MVNO-style experimentation to a four-incumbent consortium bidding for MSS spectrum is a structural bet that the handover between satellite and terrestrial, and the handset, will be the choke point — and that owning spectrum is cheaper than renting someone else's constellation. For a €42 billion market cap incumbent, that is the difference between hedging and defending.

The newest worry: agentic AI arrives as a sector risk

For the first time in this company's recent history, an analyst asked whether agentic AI is a threat — specifically whether personal AI agents shopping on price will compress telecom tariffs. Christel Heydemann's answer is worth quoting at length because it is the first structured defence of the telco business model against the application layer:

when it comes to the fear of automatic, I would say, price decrease by customers. First of all, this is something that we don't think comes as a news because we already have very transparent pricing ecosystem in Europe.
Christel Heydemann, CEO · 2026-10-27

The argument: European telecoms already price transparently, already run visible low-end brands alongside premium ones, and have spent years defending churn against exactly that comparison shopping. The global tape, though, is voting on agentic AI very differently — the software layer has been de-rated hard over the past year, with system-of-record and generative-AI-capability baskets deeply negative across the 360-day window while hardware and networking names that host the workloads have been re-rated. Orange sits on the wrong side of that trade by taxonomy and the right side by cash flow. Whether the market accepts that dichotomy is the central open question for the next several quarters.

What management quietly stopped talking about

The most interesting signal is an absence. Last quarter's single highest-ranked theme was data center — the carve-out of five French sites into a 400-megawatt joint venture with a partner contributing all the equity, with Orange contributing assets and taking a co-controlled, dividend-paying stake. That project generated an entire analyst exchange in July: “we're not talking of building an AI gigafactory, but we're not excluding working in the context of our partnership that we have in place with other French players on AI” — Christel Heydemann, CEO · 2026-07-28 This quarter the data center project is not mentioned once in prepared remarks or in Q&A. Nor is the Spanish FiberCo with Vodafone, which dominated late 2024. Thematically, both have been superseded by SFR integration and the spectrum consortium. That is not necessarily bad corporate behaviour — the SFR process is the bigger, nearer event — but it is a pattern worth flagging: Orange has a habit of announcing infrastructure value-unlock vehicles with fanfare and then going quiet. It also coincides with a global tape where data center capacity and AI data center baskets have turned negative over the last 30 and 90 days, so the silence may simply be good timing.

The balance sheet is the quiet hero, and the politics are the wildcard

The rate question — French 10-year yields higher, sector derated — was answered with specifics: fixed rates on 92% of financial debt, 7.2 years average maturity, roughly EUR 14 billion issued in twelve months, and a EUR 4.1 billion bond in September that was four times oversubscribed. On whether the SFR price could be renegotiated, Heydemann acknowledged adjustment mechanisms tied to SFR's performance but said the trend is as foreseen and any renegotiation would disturb the antitrust process and the calendar — signing by end-2026, closing by end-2027. The new item here is politics. A French election next year makes the state a future shareholder in a transaction, and the political party question is now explicitly on the table: management says no party has commented negatively, and that rising rates make SFR keener to conclude, not less. That may be true, but it is also the answer a company gives when it has no control over the variable. Elsewhere the operational detail is quietly good — device-attached customers churn four points less, Orange Money merchant payments and microcredit each grew over 50% in six months, merchant payments now a genuine growth line inside Africa and Middle East, and Max-It holds 38.5 million active users, 1.8x year on year. Guidance for 2027 was refused but characterised as "more of the same." No price tape was available for this name in the context supplied, so the market's verdict on the upgrade cannot be read here — only the fact that the upgrade happened, and that the questions it fielded were about a world Orange does not fully control.