Open-source AI rivals close gap, CEO billions questioned as Qwen models lead benchmarks
Researcher argues open-source models like Qwen now rival closed-source, making massive AI investments questionable.
A viral Reddit post by user Tree8282 argues that CEO spending billions on AI is unjustified. The researcher notes that after four years of hype, even laypeople agree AGI isn't imminent. More critically, open-source models have closed the gap with proprietary systems. For example, Alibaba's Qwen models are now the benchmark of choice in the research community, and companies like GLM (China), Kimi (Moonshot AI), and Meta all compete on leaderboards. This democratization means no single closed-source model holds a multi-year advantage anymore.
The post makes a stark financial argument: a new entrant could spend ~$1 billion to poach top talent from Anthropic or OpenAI and produce a competitive model within a year. Compare that to the $100B+ already funneled into leading AI startups. The implication is that the marginal value of additional R&D spending is collapsing. While the post doesn't account for infrastructure costs (e.g., data centers), it challenges the conventional wisdom that AI requires endless capital—suggesting the market may be overvaluing proprietary research in an era of accessible open-source alternatives.
- Open-source models like Alibaba's Qwen now serve as the primary benchmarks for AI research, displacing older closed-source models.
- A new company could replicate top-tier AI capability by spending ~$1B on talent, compared to $100B+ spent by current leaders.
- The gap between open-source and closed-source AI has narrowed to less than a year, undermining the need for massive proprietary R&D investments.
Why It Matters
Challenges the trillion-dollar AI investment thesis as open-source models commoditize cutting-edge capabilities, risking overvaluation of AI startups.