Mozilla building an AI ‘rebel alliance’ to take on OpenAI, Anthropic

From his small, snow-covered farm outside Toronto, home to cats and a dog, and soon some donkeys, Mark Surman has been laying the groundwork for a fierce battle with the world’s leading artificial intelligence companies, located about 2,300 miles away in the San Francisco area.

The bespectacled 56-year-old is president of Mozilla, a nonprofit organization best known for its Firefox browser and a pledge to keep the internet open and accessible to all. Having taken on Microsoft in the browser market in the early 2000s, and Apple and Google in the years that followed, Mozilla is right at home playing the role of underdog.

These days, Surman is preoccupied with the tech industry’s influence over the next big thing: AI. And it’s too big of a challenge for Mozilla to tackle on its own.

Surman is building what he’s described as “a rebel alliance of sorts,” using a phrase that’s long been part of Mozilla’s lexicon. In this case, the alliance is a loose network of tech startups, developers and public interest technologists committed to making AI more open and trustworthy and to checking the power of industry heavyweights like OpenAI and Anthropic.

“It’s that spirit that a bunch of people are banding together to create something good in the world and take on this thing that threatens us,” Surman told CNBC in an interview. “It’s super corny, but people totally get it.”

In practice, Mozilla is focused on deploying its roughly $1.4 billion worth of reserves to support “mission driven” tech businesses and nonprofits, including its own, according to a report the organization released on Tuesday. It’s pursuing investments that promote AI transparency, and can potentially act as a counterforce to companies that are growing at historic rates with limited guardrails.

Financially, Mozilla is at a massive disadvantage. In 2022, it launched a venture capital fund called Mozilla Ventures and pledged to invest an initial $35 million in early-stage companies. It’s now exploring raising additional funds.    

Mozilla’s cash pile is dwarfed by OpenAI, which has raised more than $60 billion from investors across the globe, and its rival Anthropic, which has raised more than $30 billion, according to PitchBook. Tech megacaps like Google and Meta are also sparing no expense, shelling out billions of dollars to hire AI researchers and tens of billions a year to build out massive data centers. 

Mozilla represents a growing swath of the AI industry that’s afraid of what OpenAI has become and the power that it now wields.

When OpenAI was founded as a nonprofit AI lab in 2015, its stated goal was to “advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return.” 

But in the decade that followed, OpenAI turned into a commercial entity with astronomical growth rates, transformed largely by the launch of ChatGPT in late 2022.

OpenAI now sports a $500 billion valuation, and completed a recapitalization in October that cemented its future as a for-profit business under the umbrella of a nonprofit. It’s a structure that resembles Mozilla, but the similarities end there.

Only a few of OpenAI’s co-founders, including CEO Sam Altman, remain at the company, and a number of early employees who left have been sharply critical of what they broadly describe as a focus on growth at the expense of safety.

Among the loudest critics is co-founder Elon Musk, who departed in 2018, started a competitor called xAI in 2023, and then sued OpenAI and Altman for alleged breach of contract and financial damages. OpenAI has dismissed Musk’s efforts as part of a “campaign of harassment,” and the case is expected to head to trial in April.

OpenAI didn’t provide a comment, and xAI returned CNBC’s request for comment with an automated response.

Anthropic was founded in 2021 by a group of former OpenAI executives and researchers who disagreed with the company’s direction. But, even as it’s taken a more pro-safety stance in AI development, Anthropic has been racing alongside AI commercially, commanding a $350 billion valuation. 

Mozilla’s uphill battle is even steeper because of the position of the Trump administration, which is determined to stay ahead of China in the global AI race and has been quick to lash out at companies, states and lawmakers that are perceived as potential threats to that agenda.

David Sacks, the venture capitalist serving as the administration’s AI and crypto czar, accused Anthropic of supporting “woke AI” in October due to its approach on regulation. President Donald Trump in December signed an executive order for a single regulatory framework for AI, establishing a litigation task force to challenge state AI laws, namely those led by Democratic lawmakers.

An Anthropic spokesperson declined to comment, but directed CNBC to a blog post from CEO Dario Amodei in October. Amodei wrote in the post that Anthropic had increased its revenue run rate from $1 billion to $7 billion in nine months, “and we’ve managed to do this while deploying AI thoughtfully and responsibly.”

Surman remains undeterred, and says Mozilla will be able to help “do for AI what we did for the web.” 

“There is an alternative that’s real and is emerging, and it’s a lot of small pieces that add up to that alternative,” Surman said. “The people in it are hungry to look where there’s weak spots in the current market and take advantage of them.”

Mozilla has long viewed itself as a rebel. 

In the 2024 “State of Mozilla” report, Surman used the phrase “rebel alliance” to describe the coalition of players that helped disrupt Microsoft’s dominance over the web. In 2020, Mozilla published a report titled “Mozilla & the Rebel Alliance,” which was dedicated to the organization’s alliance of “tens of thousands of people around the globe who believe in Mozilla.”

Even so, Surman said it took some time to convince his colleagues that the moniker applied to the AI era.

That process actually started long before generative AI took off. In 2019, Surman shifted the philanthropic and advocacy efforts of the Mozilla Foundation to focus on “trustworthy AI.”

By the spring of 2023, Mozilla had launched its venture firm and its own AI company, Mozilla.ai. The following year, Surman said Mozilla’s leadership agreed that keeping AI “trustworthy and open” was a fight worth picking. 

While its biggest priority remains growing and investing in Firefox, investing in the rebel alliance is “at the heart of who Mozilla is today,” according to the report on Tuesday. Supporting startups is central to that strategy.

Mozilla Ventures has invested in more than 55 companies to date, including dozens of AI startups, with more deals to come in 2026.

Trail, a German startup that offers an AI governance solution for regulated enterprises, raised a pre-seed round in 2024, with participation from Mozilla.

Anna Spitznagel, who co-founded the company the prior year, said Trail and Mozilla are exploring ways to collaborate more closely, like by building an open-source framework. Mozilla has supported open-source technology since its origin in 1998.

But Spitznagel isn’t completely sold on Surman’s rebel alliance concept. She said it’s a “fun analogy” and wants to be aligned with the movement to enable trustworthy AI, but also wants in on the broader AI transformation.

“Rebel is a word that for me, personally, it has the wrong association,” Spitznagel said in an interview. “I do think about [AI] a bit differently, but I also want to be part of the revolution that actually enables us to deploy AI and not hinder it.”

Tony Salomone and Ali Asaria, co-founders of Mozilla portfolio company Transformer Lab, said they’re similarly on the fence.

“I’m not gonna lie, I sometimes talk that way to get people kind of excited or engaged in our way of thinking,” Salomone said.

Founded in 2024, Transformer Lab is building open-source tools that developers can use to build, train and evaluate advanced AI models. The company has yet to publicly disclose any funding and, as of November, had fewer than 10 employees, mostly based in Canada. 

Asaria said rebel alliance isn’t a term he’s used, but that there is an ecosystem of smaller AI companies that keep in touch and regularly cross paths at conferences and other events. 

“There’s definitely a group of folks who are interested in this idea of trying to be sustainable companies that can have an impact on the industry and have an appreciation for AI, but don’t want to see just a few big companies win,” Asaria said.

When it comes to the big companies in AI, Surman cautioned that a “winner-takes-all” mentality still lurks behind their open-source efforts. He said their contributions to open-source communities are welcome, but that those same companies will “eat you if you’re not careful.”

It’s an issue that resonates with Oumi CEO Manos Koukoumidis. Backed by Mozilla, Oumi operates an open-source platform that researchers and engineers can use to train, fine-tune, evaluate and deploy AI models. Koukoumidis previously spent around a decade working in AI at Microsoft, Facebook and most recently Google, where he became disillusioned with the future he was building. 

While all the big tech companies contribute to a variety of open-source projects, some of which they manage, Koukoumidis said the bigger objective at the “tech mammoths” is dominance. In terms of safety, he is “very confident that they’re taking a lot of shortcuts.”

Koukoumidis and Surman agree that a much larger community of researchers and entrepreneurs should be collaborating to advance AI, which is one of the goals of Oumi. 

“Even the couple thousand people that are at OpenAI, Anthropic or anywhere else, because they’re operating in a silo, they’re not enough to advance this technology sufficiently, safely, cost efficiently, sustainably,” Koukoumidis said in an interview. “What’s happening right now, it’s complete insanity. We’re wasting billions, tens of billions, hundreds of billions.”

But Koukoumidis knows that abandoning a high-paying job at a place like Google has its drawbacks. He has substantially fewer resources at his disposal, and said his decision to leave the company was “intimidating.” 

When the Transformer Lab team set out to raise funding in Silicon Valley and Canada, they were repeatedly told that it was going to be “technically impossible” for them to compete. 

“When you enter into the space of AI as a new startup, it’s scary, because these few companies control so much more than just the intellectual property,” Asaria said. They control funding and access to infrastructure, making it “very hard to just walk into the space without starting with $100 million or a billion dollars,” he said.

Surman acknowledges he has to play the long game.

By 2028, he wants Mozilla to be funding a growing open-source AI ecosystem that’s on its way to becoming “mainstream” for developers. And he’s determined to prove that Mozilla’s approach is economically viable.

Mozilla is targeting a series of financial metrics over the next few years, including 20% annual growth in non-search revenue, according to a November report.

“For many people, the idea that open-source AI can win, or this rebel alliance, that those players can actually take a piece of the market, they find it hard to believe,” Surman said. “But there’s a bunch of trends that are underway.”

WATCH: AI honeymoon is over and this will be its hardest year yet, says Deutsche Bank’s Adrian Cox