Why open source AI is worth fighting for

image via TechRadar

August 14, 2026, 9:26 AM

• •The piece is an opinion article in TechRadar Pro Perspectives written by Featherless.ai CEO and co-founder Eugene Cheah.
• •Cheah says dependence on proprietary cloud AI providers creates strategic, security and accounting liabilities for enterprises.
• •He cites Berkeley-linked research claiming enterprise compute costs can fall from about $3,000 to $31 when moving from proprietary APIs to open-source.
• •The article says open-source models are close to leading proprietary systems on LMSYS leaderboard performance.
• •Cheah argues regulatory and geopolitical tensions, including data protection concerns in Germany, strengthen the case for sovereign AI deployments.
• •He contends most companies need adaptable task-specific models they control locally rather than access to the most advanced closed systems.

Eugene Cheah argues that enterprises are moving away from proprietary AI services toward open-source models because of cost, security and control concerns. He cites claims that switching from proprietary APIs to open-source systems can sharply reduce computing costs while narrowing the performance gap with top closed models. The article also frames open AI as a geopolitical and regulatory issue, especially for organizations wary of relying on foreign hyperscalers for sensitive workloads. Cheah says businesses increasingly want sovereign AI systems they can run, adapt and govern within their own jurisdictions.

Read original article