Nonprofit Current AI Raises $400 Million to Build a Government-Backed Alternative to Big Tech's AI Models

A nonprofit with a French government check for $100 million wants to build what it calls the World Wide Web of AI. It's called Current AI, and it just spent the past two months rolling out grants, tools, and a language device for rural India.

The group launched in February 2025 at the Paris AI Action Summit, founded and chaired by Martin Tisné, who also runs AI Collaborative. According to Crypto Briefing, Current AI has secured more than $400 million in funding commitments and is targeting $2.5 billion over five years.

The backer list is not exactly a scrappy underdog roster. The French government put in $100 million. Google, Salesforce, DeepMind, the Ford Foundation, and the MacArthur Foundation are also in, according to TechCrunch. UN Secretary-General António Guterres and European Commission President Ursula von der Leyen have both put out supportive statements, per Crypto Briefing.

Current AI CEO Ayah Bdeir, who joined in January after leading Mozilla's AI strategy and founded the STEM company littleBits before selling it to Sphero in 2019, drew a distinction for TechCrunch: these funders are not investors. "They're funders," she said. No equity, no return expected.

What They've Actually Built

This isn't just a mission statement with a big bank account behind it. In February, at the India AI Summit, Current AI teamed up with Bhashini, the Indian government's AI language division, to launch Suno Sutra, a pocket-sized offline device running AI in 22 Indian languages. TechCrunch reports it's open-sourced so developers can build on it.

The timeline since has moved fast. On June 19, 2026, Current AI announced $3.2 million in grants for cultural preservation projects spanning sub-Saharan Africa, the Arab region, and the Brazilian Amazon, according to Crypto Briefing. On July 1, the group released "Gap Map v0.1," a diagnostic tool meant to flag holes in the open-source AI stack. On July 9, it launched something called an "AI Potluck" to pull in community contributions to public AI infrastructure. Then, at the AI for Good Summit in Geneva, Current AI launched an open-source chatbot, per TechCrunch.

The Actual Argument

Bdeir's case is straightforward: every major AI system today, OpenAI, Google, Anthropic, is owned by a private company optimizing for market share and profit. If AI is going to reshape daily life the way she and Tisné argue it will, there needs to be a public option nobody can lock down or paywall.

The language angle is where this gets concrete instead of abstract. English dominates the large language models that most people interact with. Half the world's spoken languages are already at risk of extinction, and TechCrunch notes that when a model can't speak a language, communities lose more than convenience, they lose a vehicle for tradition and identity.

Bdeir pointed to a specific example of what she sees as extraction rather than partnership: for Indigenous languages, missionary Bible translations have ended up as AI training data before the communities who speak those languages set any rules for how their language gets used. Big Tech's multilingual expansion happens on the company's terms, not the community's.

The Obvious Tension Worth Naming

Google is both a funder of Current AI and one of the exact companies whose "walled garden" approach the nonprofit says it's trying to counter, per Crypto Briefing's framing. Salesforce and DeepMind, also funders, are commercial AI players themselves. That doesn't automatically make the project hollow. Plenty of nonprofits take money from industry players with mixed motives, but it does mean Current AI's independence from the incentives of Big Tech isn't as clean as the pitch suggests. No source here reports any governance mechanism that would prevent funders from steering priorities.

A fair skeptic would also note the funding gap: $400 million has been committed against a $2.5 billion five-year target. Crypto Briefing frames the $400 million as a war chest rivaling decentralized AI protocol treasuries, but notes that whether the full $2.5 billion target actually materializes is still an open question, and either figure is a fraction of what a single major AI lab spends on compute in a year.

What Happens Next

Current AI has no blockchain, no token, and no crypto component, according to Crypto Briefing, despite philosophical overlap with decentralized-tech arguments about resisting corporate gatekeeping. It remains, structurally, a traditional nonprofit betting that open licensing and multilingual access can out-compete trillion-dollar labs on relevance to the people whose first language isn't English.

A device running 22 languages offline in rural India is a real, usable product today. Whether $400 million, or even the targeted $2.5 billion, is enough to build comparable infrastructure across the thousands of languages Bdeir says are being left behind is something no source has yet answered.