{"id":84191,"date":"2026-01-14T09:00:00","date_gmt":"2026-01-14T17:00:00","guid":{"rendered":"https:\/\/blog.mozilla.org\/?p=84191"},"modified":"2026-01-14T08:06:22","modified_gmt":"2026-01-14T16:06:22","slug":"mozilla-ventures-2025-portfolio-convening","status":"publish","type":"post","link":"https:\/\/blog.mozilla.org\/en\/mozilla\/mozilla-ventures-2025-portfolio-convening\/","title":{"rendered":"How founders are meeting the moment: Lessons from Mozilla Ventures\u2019 2025 portfolio convening"},"content":{"rendered":"
\n
\"Mozilla<\/figure><\/div>\n\n\n

At Mozilla, we\u2019ve long believed that technology can be built differently \u2014 not only more openly, but more responsibly, more inclusively, and more in service of the people who rely on it. As AI reshapes nearly every layer of the internet, those values are being tested in real time.<\/p>\n\n\n\n

Our 2025 Mozilla Ventures Portfolio Convening Report<\/strong><\/a> <\/strong>captures how a new generation of founders is meeting that moment.<\/p>\n\n\n\n

At the Mozilla Festival 2025 in Barcelona, from Nov. 7\u20139, we brought together 50 founders from 30 companies across our portfolio to grapple with some of the most pressing questions in technology today: How do we build AI that is trustworthy and governable? How do we protect privacy at scale? What does \u201cbetter social\u201d look like after the age of the global feed? And how do we ensure that the future of technology is shaped by people and communities far beyond today\u2019s centers of power?<\/p>\n\n\n\n

Over three days of panels, talks, and hands-on sessions, founders shared not just what they\u2019re building, but what they\u2019re learning as they push into new terrain. What emerged<\/a> is a vivid snapshot of where the industry is heading \u2014 and the hard choices required to get there.<\/p>\n\n\n\n

Open source as strategy, not slogan<\/strong><\/h2>\n\n\n\n

A major theme emerging across conversations with our founders was that open source<\/a> is no longer a \u201cnice to have.\u201d It\u2019s the backbone of trust, adoption, and long\u2011term resilience in AI, and a critical pillar for the startup ecosystem. But these founders aren\u2019t na\u00efve about the challenges. Training frontier\u2011scale models costs staggering sums, and the gravitational pull of a few dominant labs is real. Yet companies like Union.ai, Jozu, and Oumi show that openness can still be a moat \u2014 if it\u2019s treated as a design choice, not a marketing flourish.<\/p>\n\n\n\n

Their message is clear: open\u2011washing won\u2019t cut it<\/strong>. True openness means clarity about what\u2019s shared \u2014weights, data, governance, standards \u2014 and why. It means building communities that outlast any single company. And it means choosing investors who understand that open\u2011source flywheels take time to spin up.<\/p>\n\n\n\n

Community as the real competitive edge<\/strong><\/h2>\n\n\n\n

Across November\u2019s sessions, founders returned to a simple truth: community is the moat<\/strong>. Flyte\u2019s growth into a Linux Foundation project, Jozu\u2019s push for open packaging standards, and Lelapa\u2019s community\u2011governed language datasets all demonstrate that the most durable advantage isn\u2019t proprietary code \u2014 it\u2019s shared infrastructure that people trust.<\/p>\n\n\n\n

Communities harden technology, surface edge cases, and create the kind of inertia that keeps systems in place long after competitors appear. But they also require care: documentation, governance, contributor experience, and transparency. As one founder put it, \u201cYou can\u2019t build community overnight. It\u2019s years of nurturing.\u201d<\/p>\n\n\n\n

Ethics as infrastructure<\/strong><\/h2>\n\n\n\n

One of the most powerful threads came from Lelapa AI, which reframes data not as raw material to be mined but as cultural property. Their licensing model, inspired by M\u0101ori data sovereignty, ensures that African languages \u2014 and the communities behind them \u2014 benefit from the value they create. This is openness with accountability, a model that challenges extractive norms and points toward a more equitable AI ecosystem.<\/p>\n\n\n\n

It\u2019s a reminder that ethical design isn\u2019t a layer on top of technology \u2014 it\u2019s part of the architecture<\/strong>.<\/p>\n\n\n\n

The real competitor: fear<\/strong><\/h2>\n\n\n\n

Founders spoke candidly about the biggest barrier to adoption: fear. Enterprises default to hyperscalers because no one gets fired for choosing the biggest vendor. Overcoming that inertia requires more than values. It requires reliability, security features, SSO, RBAC, audit logs \u2014 the \u201cboring\u201d but essential capabilities that make open systems viable in real organizations.<\/p>\n\n\n\n

In other words, trust is built not only through ideals but through operational excellence<\/strong>.<\/p>\n\n\n\n

A blueprint for builders<\/strong><\/h2>\n\n\n\n

Across all 16 essays, a blueprint started to emerge for founders and startups committed to building responsible technology and open source AI: <\/p>\n\n\n\n