Sopra Steria Discloses 19.7 Million Shares—and 25.5 Million Votes. The Difference Tells You Who's in Charge

Every month, Sopra Steria — the French IT services and digital-consulting group — files a two-line disclosure with the market regulator, and every month it reports a number that looks like a typo. As of August 31, 2026, the group said it had 19,689,538 shares in issue but 25,488,753 "theoretical" voting rights. More votes than shares. A U.S. reader stares at the mismatch the way a detective stares at a number that refuses to reconcile: something is being counted twice.

The explanation is legal rather than scandalous, but it is the single most informative thing in the filing. In France a shareholder who holds shares in registered form for at least two years earns double voting rights — one vote per share, plus one more. Sopra Steria's disclosure is largely a census of how many of its 19.7 million shares carry that bonus. The extra 5.8 million votes tell you, in round numbers, which roughly three in ten shareholders are long-term, registered, and influential.

The filing also carries a second, quieter signal. The company lists 25,488,753 theoretical votes but only 25,128,416 votes that can actually be exercised. Those 360,000 or so votes belong to treasury shares the company holds and, under French rules, cannot vote. That is roughly 1.8% of the share capital sitting idle in the group's own hands — the residue of a capital-return program that, in April 2026, culminated in the retirement of the shares it had bought back.

None of this is the smoke-alarm material a forensic reader hunts for. What the numbers establish is a control structure, and control is where a shareholder's practical power lives.

Who owns the double votes

The share-capital breakdown the group publishes does the arithmetic out loud. Sopra GMT, the vehicle through which the founding Pasquier family and co-founders hold the company, owns 20.5% of the shares but carries 29.9% of the theoretical votes. Founders and management add 2.6% of shares and 3.9% of votes; employee share plans hold 6.2% of shares and 8.5% of votes. Add the three and the insider-aligned bloc controls roughly 42% of the vote — twenty percent higher than its share of the capital — because the long-held, registered stock is precisely where the double voting accrues.

The other side of the ledger is just as pointed. Public holders — everyone else, including any U.S. investor buying the ADR on the U.S. OTC market — own 68.6% of the shares but swing only 56.1% of the theoretical votes. The utility of a loyalty-vote structure is that it concentrates say in the hands of holders who keep the stock, and here the people who keep it are the founding family, management, and the staff: the bloc that also happens to run the board.

This is not fraud, and it is not even unusual. Two dozen French large-caps use loyalty voting. The correct label is governance structure: a durable, disclosed arrangement under which the people who elected the board hold a voting cushion over the people who merely supplied the capital. A minority investor pays the same economics and faces a board that is not, practically, accountable to them.

Why the timing matters

The disclosure at the end of August lands right after the company's best quarter in years. In late July Sopra Steria reported first-half 2026 revenue of €2,958.9 million, up 4.1% in total and 3.0% organically, with second-quarter organic growth accelerating to 5.3% from 4.4% in the first. Operating margin on business activity widened to 9.6% from 9.2%. On that print the shares jumped about 15%.

The headline reason for the move, and the one most markets saw, was guidance: management raised its full-year 2026 organic-growth target to 2.0%–2.5% from 1.0%–2.0%, and the math is more flattering than the headline because roughly two points of drag come from winding down the low-margin SFT program — strip that out and the underlying target is 4.0%–4.5%. Demand is concentrated where Europe is spending: public sector, defense, aerospace and space, and financial services, which together now supply two-thirds of group revenue.

Seen through the share-count lens, the investor case has a second, quieter leg. A business returning money to shareholders via a dividend and buybacks, one that now retires the repurchased shares rather than leaving them in treasury, delivers per-share growth even when revenue grows slowly. Fewer shares, same earnings, more value per holder. The valuation leaves room for that arithmetic to matter: in early September the group traded at a trailing price-to-earnings ratio near 10, with a forward multiple in single digits.

The parts that do not reconcile

But the same documents that make the bull case legible carry the caveats, and a careful reader should hold both.

The first is geography and mix concentration. Just over a third of employees are in France, and roughly half of revenue originates there; the growth engine is heavily a story of national governments and defense budgets, which are real but which concentrate risk in a handful of spending programs and political cycles.

The second is the SFT wind-down the growth math leans on. Management presents the drag as temporary, and the underlying number is stronger — but it is a business being exited, and the guidance explicitly banks on the accounting view the company itself chooses to foreground.

The third is balance-sheet headroom and cash timing. Net financial debt was €618.7 million at June 30, leverage at 1.1 times EBITDA and gearing improved to 29% from 34% — a manageable position, but not a cash-rich one. And first-half free cash flow was negative €143.6 million. Management calls that seasonal — the group typically generates its cash in the second half — and it is a real phenomenon in this industry, not a red flag on its own. It is a reminder that the buyback-and-dividend machine is funded by money that has to show up in the second half to validate the first-half story.

The invoice attached to the double vote

There is no fraud here and no restatement looming. The monthly disclosure is a routine, accurate census. What it portrays is a real, growing, moderately priced European IT services company whose shareholder-return machinery quietly compounds value through buybacks and a dividend, and whose governance equally quietly under-weights the votes of the outside capital that funds it.

The cost of the structure is not a line in any income statement. It is the premium a minority investor implicitly pays for owning economics they cannot steer: roughly 42% of the voting power sits with the founding and employee bloc on about 29% of the capital, and that cushion means board decisions — on pay, on acquisitions, on the pace of return of cash — are made among the long-term faithful, not negotiated with a spread-out base.

For a U.S. retail shareholder in particular, the practical consequence is to buy the company for the operational story and the buyback math, and to treat governance as a standing discount rather than a defect that will be fixed. The double-vote structure is not going away; it is the point. The next documents that matter are ordinary ones: the second-half cash-flow reports that confirm the seasonal collection actually lands, and the next monthly disclosure showing whether the share count keeps shrinking. Both are already scheduled. The status is Level One — an anomaly fully explained by law, not evidence of anything being hidden. The control structure, once decoded, is simply the price of admission this company asks of everyone who buys without first reading the vote count.