Open-source language and generative models are increasingly used by enterprises, governments and startups, drawn by the promise of lower costs, greater control and more flexibility. On TechCrunch’s Equity podcast, Rebecca Bellan spoke with Irene Solaiman, Chief Policy Officer and SVP of Strategy at Hugging Face, about what organizations should consider before building on open models.
Why many are switching to open-source
Solaiman noted that startups, enterprises and public institutions often choose open models to reduce costs and retain more operational control. Open models can offer more flexible deployment options and lower dependence on pay-per-use pricing, which may be advantageous for long-running, high-volume use cases. She also cautioned that using the largest-capacity models is not always sensible when the cost-to-benefit ratio or operational complexity is unfavorable.
Who captures value when models are free
The conversation addressed who benefits financially when models themselves are freely available. According to Solaiman, open models are useful, but much of the captured value comes from services, customization, infrastructure and managed offerings. Integrators, cloud providers and enterprise implementation teams are among the actors who can build revenue streams around open models by packaging and operating them for customers.
China’s rise and whether the U.S. needs deep-pocketed players
Bellan and Solaiman discussed the competitive implications of Chinese open models gaining traction. Solaiman argued that if leading global models are shaped primarily by Chinese development, there could be strategic and economic consequences. Meeting that challenge, she suggested, may require larger financial commitments from Western actors — for example companies like Nvidia — able to sponsor and sustain an open but globally competitive frontier model.
Security versus openness: attacks raise hard questions
They also covered recent attacks on open AI platforms and the tension these incidents expose: how can an ecosystem be kept secure without shutting it down? Solaiman framed this as a core dilemma for open systems, where security controls and access restrictions can conflict with the innovation and broad applicability that openness enables.
Practical considerations for builders
Solaiman offered practical advice to startups and organizations planning to build on open models: choose the right tools and infrastructure, carefully weigh cost and benefit trade-offs, and recognize that monetization often comes not from licensing the model itself but from the services and integrations built around it.
Where to listen
The full interview appears on TechCrunch’s Equity podcast, available on YouTube, Apple Podcasts, Overcast and Spotify. The program can also be followed on X and Threads at @EquityPod.
