• Open-weight AI models like Kimi K3, Qwen, and Llama are expected to lower costs and accelerate adoption, not undermine market leaders.
  • Morgan Stanley highlights the Jevons Paradox: as AI becomes cheaper, overall demand for compute could rise.
  • Nvidia (NVDA) is positioned as the biggest winner, with sustained demand for AI infrastructure regardless of model architecture.

The New AI Paradigm

Morgan Stanley is challenging the fear that open-weight AI models will erode the dominance of proprietary leaders. Instead, the bank argues that the rise of models such as Kimi K3, Qwen, and Llama will expand the AI market by making the technology more affordable and accessible. The key driver, according to the analysts, is the Jevons Paradox: as the cost of AI drops, usage and demand for compute will likely surge, benefitting the entire ecosystem.

“We see a future where open and closed models coexist, with enterprises leveraging both to optimize performance and cost,” said the Morgan Stanley note, which was shared with clients earlier this week. The bank’s survey data indicates that 63% of enterprises already run a mix of open and closed models, a trend expected to accelerate as open-weight options become more sophisticated.

Nvidia’s Central Role

Regardless of which AI architecture wins, Morgan Stanley asserts that Nvidia stands to benefit most. The demand for AI computing infrastructure is projected to grow as cheaper models spur broader deployment, particularly in inference-heavy workloads. This could lead to increased spending on GPUs, cloud capacity, memory, and networking—all areas where Nvidia has a commanding presence.

“The compute requirements for AI are far from saturated,” noted a senior analyst at the bank. "As the cost per unit of intelligence falls, we anticipate a dramatic expansion in use cases, which will drive a corresponding increase in infrastructure investment."

Implications for the Industry

The report suggests that the narrative of open-weight models as a threat to incumbents is overblown. Instead, the proliferation of accessible AI is likely to fuel a broader shift in spending toward hardware and infrastructure, with sovereign AI initiatives also contributing to demand. While margins for closed-model leaders may face pressure in the short term, the overall market expansion could compensate for volume.

Enterprises, meanwhile, are increasingly adopting hybrid strategies, choosing open models for flexibility and cost efficiency while leveraging closed models for performance and support. This dual approach is expected to raise total AI-related capital expenditures over time, a positive signal for the entire supply chain.

Looking Ahead

As the debate over AI safety and governance continues, open-weight models are drawing regulatory attention in the U.S. and beyond. Morgan Stanley’s analysis suggests that policy decisions will play a critical role in shaping the market’s trajectory, but the direction of travel is clear: AI is becoming more democratized, and the infrastructure providers will reap the rewards.

“The future of AI is not a zero-sum game,” the note concluded. "The expansion of the market will benefit everyone, from model developers to chipmakers."

For now, investors appear to be taking the optimistic view, with Nvidia’s stock holding steady amid the ongoing AI frenzy. The bank’s outlook offers a counterpoint to concerns that open-source models could commoditize the industry, suggesting instead that they will unlock new opportunities for growth.