A10 Networks, the San Jose-based network and application security company listed on the New York Stock Exchange, said on June 15, 2026 that it had acquired TrojAI, an AI security startup based in Saint John, New Brunswick. The deal closed on the day it was announced.
Financial terms were not disclosed. A10 said it did not expect the purchase to have a material impact on its financial results for fiscal year 2026.
TrojAI's 22 employees will stay in Saint John, according to the Atlantic Canadian startup news site Entrevestor, and the company is likely to hire more people there over time. A10 plans to combine TrojAI's software with its own hardware-based AI firewall products.
What TrojAI built
TrojAI sells tools to secure, test and govern AI applications and agent-based workflows. Its product works in two stages. During development, it red-teams models, agents and applications, attacking them to find weaknesses before they are released. Once they are deployed, it provides runtime protection that acts as an AI firewall and blocks threats in real time.
In a blog post about the deal, A10 said TrojAI's product also supports agent protocols, including native compatibility with the Model Context Protocol (MCP), and offers browser-based endpoint protection. A10 said the startup serves Fortune 50 customers, without naming them.
“TrojAI is a natural fit for A10, strategically and operationally.”
Why A10 is buying
A10 framed the deal around what it calls sovereign AI security: letting customers decide how and where their AI models, data and agents are protected, whether on premises, in the cloud or in hybrid setups. Its existing AI firewall runs on hardware; TrojAI adds software for red teaming and runtime defence.
"AI is changing both what enterprises build and the attack surface they have to defend," said Dhrupad Trivedi, A10's president and chief executive. He described TrojAI as "a natural fit for A10, strategically and operationally."
In the blog post, A10 executive Jamison Utter wrote that the deal was "not a change in direction" but an acceleration of a strategy the company had already been building. He said findings from TrojAI's red-team testing would feed into A10's guardrail models in near real time. The combined portfolio now spans application delivery, DDoS protection, application and API security, and AI security.
The founders and the funding
TrojAI was founded by James Stewart and Stephen Goddard, who had earlier sold their previous company, EhEye, to Toronto-based Patriot One Technologies for more than $3.2 million in stock, Entrevestor reported. Stewart led TrojAI as chief executive until early 2024, when Lee Weiner, a former product and technology head at Rapid7, took over as CEO and Stewart became chief technology officer. Goddard is chief operating officer.
The company raised C$3 million (about $2.3 million) in 2022 in a round co-led by Flying Fish Ventures. In April 2024 it closed a further $5.75 million (C$7.77 million) of seed funding, again led by Flying Fish Ventures. Techstars and Halifax-based Build Ventures added to their stakes, the New Brunswick Innovation Foundation put in $500,000, and Alteryx Ventures and Flybridge Capital Partners joined as new investors. At the time, TrojAI said it would open an office in Boston.
"Enterprises and public-sector organizations are adopting AI at an unprecedented pace," Weiner said in the acquisition announcement.
How the deal came together
According to Entrevestor, A10 approached TrojAI about six months before the announcement, and the two sides then began a formal sale process. Stewart told the publication that he and Goddard were impressed by A10 as talks went on, and that the teams found a good match in technology, culture and vision.
"We're all staying here in New Brunswick," Stewart said. Goddard said the deal meant the team could continue its work in Atlantic Canada together.
A10 had a market value of about $2.4 billion (C$3.4 billion) at the time, and its shares rose about 4.7% on the day of the announcement, Entrevestor reported.
A crowded field for AI security
Large security and networking vendors have been buying startups that test and protect AI models and agents, as companies put generative AI into production and connect agents to internal tools through protocols such as MCP. A10 said it expected the combined product to help customers run AI wherever their data sits over the next two to five years, particularly in organisations with strict data-sovereignty requirements.
