Phoenix New Media's Fastest-Growing Business Is the One Nobody Will Talk About
Q3 revenue rose 8.3% — but advertising fell, digital reading carried the quarter, and management spent the call narrating a segment that is still too small to matter
FENG · Earnings Call · 2026-11-10
A growth headline that hides a swap
Phoenix New Media reported third-quarter 2026 total revenue of RMB 217.6 million, up 8.3% year-on-year. That is the number the press release leads with, and on the tape it looks like a media company finding its footing. It is not. Underneath, the engine changed hands: net advertising revenue fell to RMB 149.2 million from RMB 159.3 million a year ago, while paid services jumped 64.4% to RMB 68.4 million. The company grew despite its core advertising franchise shrinking, not because of it.
The cause of the paid-services surge is stated once, in the CFO's script, and never mentioned again by anyone on the call:
Paid services revenues were RMB 68.4 million, representing a 64.4% increase year-on-year from RMB 41.6 million, primarily driven by the revenue generated from our digital reading services offered through mini programs on third-party applications.
That is a distribution arbitrage on someone else's platform — the same mini program channel flagged as a top theme back in Q1 2025, when paid services first ranked as the company's number one keyword. It is now the entire growth story. And it is not free: total operating expenses rose 18% to RMB 128.6 million, driven by sales and marketing for exactly that business. Gross margin improved sharply to 57.6% from 47.6%, operating loss narrowed to RMB 3.2 million from RMB 13.3 million — and yet net loss attributable to ifeng widened to RMB 24.3 million from RMB 4.9 million. No one on the call explains that gap. The loss from operations line, a recurring keyword in this company's vocabulary for years, is doing the opposite of the bottom line.
The advertising reversal is real, and it is a full reversal
Three quarters ago, management was listing categories that grew. In May 2026: “Areas like liquor, Internet services, automotive, home appliances, finance and retail all grew year-on-year.” — Xiaojing Lu, Executive (likely CEO or senior management) · 2026-05-13 In August, the tone shifted to "mixed" and management argued the content-led model was “more of a structural shift rather than just a short-term thing.” — Xiaojing Lu, Senior Management / Possibly CFO or Head of Advertising · 2026-08-11
This quarter, the same analyst asked the same question about advertising, and the answer inverted the category list she was given in May: “This quarter, revenue from home appliances, e-commerce and automotive clients was lower. Revenue from liquor and food and beverage clients grew.” — Xiaojing Lu, Management · 2026-11-10 Two of the three growth pillars from Q1 are now declining; liquor, which was the drag in Q2 on contract-renewal timing, is back to growing. That is not a structural shift — it is quarterly churn inside a soft ad pool, and the analyst's persistence (identical question, three calls running) suggests the street treats it as unresolved rather than explained.
The offset is a genuine, narrow win in tech clients. Revenue from AI and robotics advertisers in the first nine months of 2026 exceeded “more than 3x the amount generated in the full year of 2025, with most of the revenue coming from original content services.” — Xiaojing Lu, Management · 2026-11-10 It is real, it is company-unique, and it is almost certainly small in absolute terms — a triple off a tiny base against a RMB 149 million ad line.
AI as a traffic window, not a revenue line
The most forward-looking thing said on the call was not about advertisers at all:
“We're also exploring how to make our content more accessible to AI models and platforms as we prepare for changes in how people find and consume information, and as AI models evolve to become the traffic window.” — Li Qi, Management · 2026-11-10
This is AI models entering the company's top keywords for the first time — a defensive posture, not an offensive one. Phoenix is a distribution business being asked what happens when the search box stops sending clicks. That fear is now global: the recent reporter tape is littered with companies narrating AI-induced traffic disruption and personal AI agents displacing the front door, and Chinese peer AI related semiconductor names are pitching entirely different parts of the same wave. Phoenix's answer — make content machine-readable, monetize brand-sponsored brand-sponsored documentaries and industry event coverage — keeps the revenue model firmly human. The CEO himself hedges it as exploring AI measured pace. Nothing here is a monetization plan.
What quietly fell off the list
For years this name ran on a vocabulary of quality content, brand influence and public-sector credibility — the former two were among the single biggest momentum gainers in the company's own keyword history. Neither appears in the current top-30. Nor does the public-sector push that dominated late 2025 (Shanxi tourism, government partnerships) or the overseas-campaign narrative of mid-2026. What replaced them is operational and inward: original content as a service deliverable, specialized content sold by vertical, data analysis and ground reporting as production inputs.
That is a meaningful tell. When a media company stops talking about influence and starts talking about content-as-deliverable, it has accepted that its brand premium is no longer the pitch — the client brief is. Brand partners still get documentaries and forum impressions (800 million for the Greater Bay Area Finance Forum), but the language has shifted from persuasion to execution.
The balance sheet is the actual thesis
Strip out the narrative and Phoenix is a sub-scale Chinese content shop sitting on RMB 935.4 million — roughly USD 139.5 million — in cash, term deposits, short-term investments and restricted cash, against a market capitalization around USD 349 million. The advertising business is shrinking, the paid-reading business is buying growth with marketing spend, and net loss is widening. Nothing in this call changed the arithmetic; it merely clarified which line is now doing the work. The uncomfortable question no one asked: if digital reading is the growth engine, why does it get one sentence of prepared remarks and zero strategic airtime, while a 3x-off-a-small-base AI/robotics advertising win gets three paragraphs? The answer is usually that the growth line is rented — distribution on third-party platforms with rising acquisition costs — and management knows it.
Bottom line: an 8.3% revenue print that masks a shrinking ad franchise, a rented paid-services engine, and a widening net loss. The AI applications story is the most interesting thing management said and the least material thing it has. Watch the digital reading marketing line, not the AI headlines.