For stock investors, analyst research reports offer authoritative, professional, timely and comprehensive insight to help uncover potential thematic opportunities. AI security is essentially about installing a safety valve on AI itself.
On one side there is defense: large models being attacked through prompts, training data being poisoned, agents being induced to misuse tools, and generated content crossing compliance red lines. On the other side there is use: letting large models in turn help security teams with alert analysis, threat hunting and automated response.
In the past, security companies competed on firewalls and signature databases. Now they compete on who can bring AI into the defense process first. In January 2026, the revised Cybersecurity Law officially took effect, writing AI security provisions into law for the first time. In September, during Cybersecurity Awareness Week, the AI Security Governance Framework 3.0 was released. With both policy and demand pushing it forward, AI security has shifted from optional to mandatory.
But AI security is not as visible or tangible as a firewall. The hardest evidence for judging whether a company is truly placing a bet lies in its financial statements: how much it spends on R&D and how aggressively it invests. This issue focuses on the AI security industry chain and selects the 10 cybersecurity companies with the highest R&D spending, to see who is truly putting real money into technology, for further reference and research. Note: the following content absolutely does not constitute any investment advice, guidance or commitment. It is for discussion and research only.
First: Sangfor Technologies (the leading security cloud player)
Concept relevance: With AI First as its R&D strategy, its security side focuses on three major directions: AI security operations, office security and protecting AI, supported by products such as AI computing power gateways and agent development platforms. R&D spending: R&D expenses of 1.027 billion yuan, accounting for 25.70% of operating revenue. Company highlights: Its application delivery product ranked first in China for four consecutive years with a 21.1% share, and its hyper-converged market share also ranks first in China. Both core businesses, security and cloud computing, are industry leaders.
Second: Qi An Xin (the sentinel for government and enterprise security)
Concept relevance: Built on its self-developed QGPT large model, it has formed a dual-wheel system of AI empowering security and AI system security, covering full lifecycle protection for large models, zero trust and AI security operations platforms. R&D spending: R&D expenses of 578 million yuan, accounting for 38.60% of operating revenue. Company highlights: Its endpoint security market share is 28.7%, ranking first in China for eight consecutive years, and its security analysis and operations rank first for six consecutive years. It was selected in all 16 secondary tracks of the Security Niu landscape map.
Third: Topsec (the veteran firewall brand)
Concept relevance: It has built a complete product matrix for large model security gateways, data monitoring and security assessment. Among its patent reserves, 371 are related to AI security, 396 to data security and 118 to intelligent computing cloud. R&D spending: R&D expenses of 320 million yuan, accounting for 58.76% of operating revenue. Company highlights: Traditional firewalls have ranked first in China by market share for 26 consecutive years, and it is also the creator of China's first commercial firewall. Revenue from AI security protection grew by more than 130% year on year.
Fourth: Venustech
Concept relevance: It jointly built a cybersecurity foundational large model with China Mobile's Jiutian, launched the new three-piece set for agent security and endpoint AIDR products, and has embedded AI Coding into its own R&D process. R&D spending: R&D expenses of 271 million yuan, accounting for 21.96% of operating revenue. Company highlights: It has ranked first in the overall data security market by market share for 10 consecutive years. As China Mobile's cybersecurity platform, its net profit attributable to shareholders in the first half grew by 132.98% year on year.
Fifth: NSFOCUS (the leading anti-leakage player)
Concept relevance: Its R&D direction is concentrated on AI security. It has completed product conversion of agent security detection and protection and large model security fence technology, and participated in formulating the Product Security Guide for Large Model Security Gateways. R&D spending: R&D expenses of 256 million yuan, accounting for 31.49% of operating revenue. Company highlights: It ranks second in China in hardware anti-DDoS share and first in data leakage prevention share. In the first half, AI security orders reached 117 million yuan, and orders signed with telecom operators grew by more than 300% year on year.
Sixth: Anheng Information (the data security steward)
Concept relevance: It released the Hengnao 4.0 security agent foundation, covering the two major directions of AI-native security and trusted data spaces. Its security large model has completed real-world testing in major event scenarios. R&D spending: R&D expenses of 209 million yuan, accounting for 28.33% of operating revenue. Company highlights: Its data security management platform and database auditing both rank first in China by market share. Revenue growth for AI-native security and trusted data spaces both exceeded 300%.
Seventh: CETC Cyberspace Security (the cryptography veteran)
Concept relevance: Its R&D investment is mainly directed toward three new directions: quantum and post-quantum cryptography, satellite internet security and artificial intelligence. Cryptographic products remain the foundation of its revenue. R&D spending: R&D expenses of 149 million yuan, accounting for 31.87% of operating revenue. Company highlights: As the leading cryptography company under China Electronics Technology Group, its cryptographic product gross margin is 54.81%. Contract liabilities increased by 59.68% in half a year, showing that customer prepayments are clearly increasing.
Eighth: Hillstone Networks (the border goalkeeper)
Concept relevance: It implements its AI strategy around two main lines: AI empowering security and protecting AI systems themselves. New products such as AI security gateways and agent security protection are planned for launch in the second half of the year. R&D spending: R&D expenses of 145 million yuan, accounting for 32.17% of operating revenue. Company highlights: Its self-developed StoneOS security operating system supports its entire product line. Its UTM share remains in China's top three, and it achieved a quarterly profit in the second quarter, with losses narrowing by 72.41% year on year.
Ninth: Zhongfu Information (the confidentiality gatekeeper)
Concept relevance: As a core supplier of confidentiality security and data security, its product system is built around classified information system protection and data security compliance scenarios. R&D spending: R&D expenses of 136 million yuan, accounting for 63.62% of operating revenue. Company highlights: It holds scarce qualifications such as classified information system integration, and entry into this track is extremely strict. Its main business gross margin rose to 78.72%, among the highest in the industry.
Tenth: Dipu Technology (the fast delivery marksman)
Concept relevance: It has a dual-line layout in cybersecurity and application delivery. Its products cover scenarios such as border protection, application security and traffic analysis, and AI capabilities have been integrated into security operations processes. R&D spending: R&D expenses of 126 million yuan, accounting for 23.41% of operating revenue. Company highlights: Its application delivery share ranks second in China, and its hardware anti-DDoS has ranked in the industry's top three for four consecutive years. In the first half, net profit attributable to shareholders was 72 million yuan, up 38.86% year on year.
The above R&D data is taken from each company's 2026 interim report and is based on the income statement basis, which may differ from the R&D investment basis in announcements. This indicator reflects technological layout, not profitability. The list only includes A-share companies whose main business is cybersecurity, with ST companies excluded. Product shares are cited from third-party reports such as IDC and CCID, and there are differences in statistical bases. They are for reference only.
Comments