--- title: "After Three Years in the Red, Zhou Hongyi Bets 360 on AI Agents" date: 2026-10-29 category: Builders site: NeuroAI canonical: https://neuroai.site/a/na-people-zhou-hongyi-360-ai language: en --- # After Three Years in the Red, Zhou Hongyi Bets 360 on AI Agents > 360 founder Zhou Hongyi pivoted his security company to "360Brain" and an "All in Agent" strategy, and returned to profit in 2025 on AI-driven growth. On a Beijing stage in September 2026, Zhou Hongyi (周鸿祎) announced he would send engineers into real shops in Beijing and Shanghai to install AI workflows by hand. The man who built China's best-known free antivirus was no longer selling software — he was delivering labour saved by agents. It is the latest turn for a founder who has spent two years reinventing a 20-year-old security company around AI. ## Key takeaways - Zhou Hongyi founded 360 (三六零), China's dominant PC-security brand, and in 2023 declared it the company's "large-model year," building the self-developed 360Brain (360智脑) model. - In 2025 he set an "All in Agent" (All in Agent) strategy, reframing 360 as an AI-and-security company rather than a pure internet-security vendor. - 360 returned to profit in 2025: revenue of RMB 8.693 billion (≈ US$1.22 billion) and net profit of RMB 263 million (≈ US$37 million), after losses in 2022–2024. - 2025 research spending was RMB 3.225 billion (≈ US$454 million), 37.11% of revenue; the company says it has invested over RMB 6 billion (≈ US$845 million) cumulatively in 360Brain. - In the first quarter of 2026, revenue was RMB 2.011 billion (≈ US$283 million) with net profit of RMB 109 million (≈ US$15 million), extending the recovery. ## From free antivirus to a model company Zhou made his name giving away security and monetizing attention. The pivot began in 2023, when he called AI the company's defining theme and launched 360Brain, a self-developed large model. By 2025 the message had sharpened into "All in Agent": build agents that do jobs — writing, office work, security operations — instead of chatbots that answer questions. The framing is deliberate. "AI has two roads," Zhou said at an industry event. "One is the 'build an atomic bomb' road the big labs take, burning compute and stacking parameters. The other is the 'make tea eggs' road — lightweight, vertical models that go to work in real scenarios and help thousands of industries." The tea-eggs line is his shorthand for 360's application-layer bet. The move is also defensive. As browsers and handsets absorbed the antivirus category, 360 needed a fresh reason for users to keep its icon on the screen — and an AI assistant baked into the browser is exactly that hook. ## The numbers behind the turn The strategy showed up in the books. According to 360's 2025 annual filing with the Shanghai Stock Exchange, revenue rose 9.37% to RMB 8.693 billion (≈ US$1.22 billion) and the company posted RMB 263 million (≈ US$37 million) in net profit attributable to shareholders — its first annual profit since 2021, after losses in 2022, 2023 and 2024. Internet value-added services, which include 360's AI office and Nano AI Search (纳米AI搜索), grew 41.38%, faster than the legacy ad business. Research intensity stayed high: RMB 3.225 billion (≈ US$454 million), or 37.11% of revenue, with the company stating it has put more than RMB 6 billion (≈ US$845 million) into 360Brain since the effort began. The first quarter of 2026 continued the trend, with revenue of RMB 2.011 billion (≈ US$283 million) and net profit of RMB 109 million (≈ US$15 million). ## What 360 actually ships The agent story is now concrete product. Beyond 360Brain and Nano AI Search, 360 released a security agent, "360 Security Lobster" (360安全龙虾), that routes tasks across more than a dozen domestic models through a coordination method the company calls CoE (专家协同, "expert collaboration"). It also launched a nano microdrama production pipeline and, in September 2026, the "Nano Work" (纳米Work) "twin-city, thousand-store" plan to put AI workflows into physical shops in Beijing's 798 art zone and Shanghai's Lujiazui. The through-line is Zhou's old instinct — meet the user where they already are, then make the AI invisible. Zhou's longer game is to fuse AI with 360's security DNA. The company's stated mission is "make the AI world safer" (让AI世界更安全更美好), and its security agents automate vulnerability discovery and threat response — a lane where a twenty-year security vendor, not a model lab, owns the data and the trust. That is why 360's AI story is told in threat dashboards and agent workflows rather than benchmark charts. According to Li An, chief scientist at BrainNet (脑机网), the 360 case shows Chinese internet incumbents are competing on the AI application layer — agents and vertical workflows — rather than on frontier model training, where capital and compute favour the cloud giants. ## Honest limitations The profit recovery is real but thin, and 360's own filings note it rests on revenue growth and lower selling expenses rather than a single AI breakthrough. Internet advertising remains the largest revenue line, so the AI businesses are still a minority of sales. The "All in Agent" positioning is a strategy statement, and the security-agent and Nano Work products are early; their contribution to profit is not yet separable in the reporting. The financial figures are drawn from 360's official exchange filings and reported by multiple financial outlets, but segment-level AI revenue is not broken out. ## What readers can do now - Buyers: test 360's Nano AI Search and agent tools as a low-cost alternative to building in-house assistants for office and security workflows. - Investors: read 360's quarterly filings for the value-added-services line as the cleanest proxy for AI-driven growth, not the headline profit. - Operators: borrow the "tea eggs" framing — pick one repetitive workflow and pilot an agent before committing to a frontier model build. --- Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-people-zhou-hongyi-360-ai Free to quote with attribution and a link to the original.