Open in interactive viewer → charts, metric popovers & call review

Xunlei: A $44 Million Market Cap, a $265 Million Cash Pile, and a 7.8% Stake Nobody Will Sell

Q3 revenue grew 12.9%, but every question on the call was about Arashi Vision — and management's answer was the same one they've given for four quarters
XNET · Earnings Call · 2026-11-12

The quarter nobody asked about

Xunlei's Q3 2026 was, operationally, an unremarkable deceleration. Revenue of $101.3 million rose 12.9% year over year, with subscription business at $44.1 million (+8.5%) and live-streaming and other services at $57.2 million (+16.6%). The CEO framed it honestly: “We delivered year-over-year revenue growth in the third quarter though at a moderated pace, as we navigated evolving market dynamics and execute strategic adjustments across our core consumer businesses.” — Luhan Tang, Investor Relations Manager · 2026-11-12 The cost line is where the story turns. Cost of revenue jumped to 43.8% of revenue from 37.6%, dragging gross margin to 55.7% from 62% a year earlier. Eric Zhou was blunt about the cause: “Live-streaming carries a lower gross margin, and now, it makes up a larger share of our total revenues.” — Eric Zhou, Chief Financial Officer · 2026-11-12 The result was an operating loss of $14.8 million versus $3.3 million of operating income a year ago — a swing that occurred before any investment mark-to-market entered the picture. Sales and marketing ran at 32.9% of revenue. Management attributes this to a deliberate user-acquisition reset and to competitive pressure in overseas audio live-streaming, where the company is spending into Latin America expansion, plus elevated acquisition costs across the Middle East and Turkey that coincided with the World Cup year. That last item cuts both ways: Hupu's advertising business got a real traffic and engagement lift, but management conceded the boost is event-driven and cyclical — they are already hunting for other products to smooth it. The global keyword engine picked up a World Cup market theme earlier this year; that tailwind is now fading for everyone.

The Arashi Vision problem swallowed the call

Then came the number that will be in every headline: a net other loss of $219.2 million, versus $547.5 million of net other income a year ago, driven by mark-to-market fair value changes on the long-term investment in Arashi Vision. Net loss from continuing operations was $235.8 million against $550.4 million of income in Q3 2025. Diluted loss per ADS was $3.74 versus $8.60 of EPS. On a non-GAAP basis — which strips the mark — the company still lost $16.4 million versus $5.6 million of income. That is accounting noise on a real asset. Xunlei owns 7.8% of a listed company, above the 5% threshold that triggers regulatory approval procedures for any sale. The stakes are not abstract: cash, equivalents, and short-term investments stood at $265.1 million against a market capitalization of roughly $44 million. The equity is being valued at a large fraction of nothing. Every question in the Q&A landed on the same thing. An analyst from Azure Crest pushed management to swap Arashi exposure for Xunlei's own paper, arguing “selling Arashi Vision and buying more XNET is even now, even though Arashi Vision decreased the price, it's highly accretive to shareholders, which would include also management” — Unknown Analyst, Analyst · 2026-11-12. The CFO's answer was noncommittal: no process has begun, the stake sits above 5%, and the timing depends on factors outside the company's control.

For the Arashi Vision divestiture, we will wait for the right time to do it. And if you look at the Arashi's price performance, it's been down quite a lot for the last several months or several quarters, and we still have high confidence in the company.

Eric Zhou, Chief Financial Officer · 2026-11-12
This is not new. In March 2026, Stonehill Capital asked the same question about the post-lockup plan. In May, management fielded the Investment Company Act question and said “we never intend to be an investment company” — Eric Zhou · 2026-05-28. In August 2026, the identical answer: “we hold more than 5% of shareholders or exactly 7.8% of Vision's equity stake... at this time, we have no specific plans for the share disposal for the time being” — Eric Zhou, Chief Financial Officer · 2026-08-13. Four quarters of shareholders asking, four quarters of patience in return. The genuinely new wrinkle is regulatory. Under the 1940 Act, investment securities cannot exceed roughly 40-45% of total assets excluding cash and government securities. Asked about the clock, Zhou said: “We haven't received any reviews or questions from relevant agencies on this issue... I believe the ratio right now is on the borderline because the Arashi's stock price has declined significantly over the last several months.” — Eric Zhou, Chief Financial Officer · 2026-11-12 Note the logic: falling Arashi shares help compliance while destroying book value. A forced-seller dynamic is lurking beneath a voluntary one. The tell is visible in the company's own keyword profile for the quarter. Valuation discourse has climbed into the top five themes — alongside stock price references tied directly to Arashi — while marketing expense sat at number three and acquisition strategy at number two. These are not operational talking points; they are the vocabulary of a company arguing about its own price.

What management quietly stopped selling

The subtler signal is the drift in the growth narrative. overseas audio live streaming was the flagship theme in prior quarters — it drove the whole "live-streaming and other" line. This quarter it reads as a headwind rather than an engine: rising user acquisition costs, intensifying competition, campaign efficiency under pressure. In Know Trend's momentum rankings it has flipped from a gainer to a decliner. The business is still growing, but it is now the source of both the revenue growth and the margin compression simultaneously. Subscription is the other soft spot. The five-year record is genuinely impressive — “Over the past 5 years, the company's subscriber business has grown steadily from about 3 million in 2020 to more than 7 million in 2025” — Eric Zhou, Chief Financial Officer · 2026-11-12 — but the reset, begun in Q2, deliberately traded initial ARPU for retention quality, and the revenue cost is hitting now. Management calls Q3 "a transitional phase" and points to improved lifetime value later. Investors have heard transition language before.

Where this leaves the equity

The capital-allocation arithmetic is the most honest signal on the call. The company spent $13.8 million of its $20 million authorization through September 30 — then exhausted the authorized share count early because the price was so low, and the Board immediately authorized another $20 million for the next twelve months. Management is voting with the only currency it controls, and voting hard, because the buyback is the highest-return use of cash available while the stock trades at a deep discount to the cash and securities on the balance sheet. The uncomfortable tension is that the same management team is content to wait indefinitely on the larger value unlock. The buyback is a slow drip; divesting Arashi is an event. Until that event happens — or until a regulatory threshold forces it — Xunlei will keep reporting enormous paper swings on an asset it controls but won't monetize, while operating a decelerating, margin-dilutive core. The repurchase program is a strong vote of confidence. It is also, at this size, an admission that the discount is too wide for the company to ignore and too stubborn for it to fully resolve.