Investors should capitalize on the booming demand for AI hardware and infrastructure by targeting NeoClouds and alternative providers like Base10 and Fireworks. As businesses increasingly shift toward cost-effective, open-source AI models, companies supplying essential data center hardware, power, and gas turbines present strong growth opportunities. Conversely, investors should exercise caution with closed frontier AI labs like OpenAI and Anthropic due to impending margin compression and pricing pressures. Monitor these traditional labs closely to see if their proprietary software wrappers, such as Cloud Code and Codex, can successfully defend their revenue streams against surging open-source alternatives.

NeoClouds / GPU Infrastructure Providers (Base10, Fireworks)

• Positioned to capture significant market economics as open-weight models proliferate and consumers shift away from exclusive reliance on frontier labs. • Benefit from strong and spectacular overall growth in the inference cloud and wider ecosystem as demand for running open-source models rises. • Capital will naturally flow to support these layers of the stack if open-weight models continue to provide valuable alternatives to closed systems.

Takeaways

• Favorable tailwinds for infrastructure providers, NeoClouds, and companies supplying hardware, power, gas turbines, or data center essentials. • Watch for increased demand in alternative cloud providers like Base10 and Fireworks as businesses opt to host open-weight models in-house.

Frontier AI Labs (OpenAI, Anthropic / Fable)

• Facing growing pricing pressure due to the rapid rise of competitive open-weight and open-source models (such as Kimi K3, Qwen, and DeepSeek). • May see their gross margins eroded as token consumption shifts toward lower-cost or open alternatives for tasks that do not require absolute frontier-level intelligence. • Relying on sticky software layers and products (such as Cloud, Cowork, Cloud Code, and Codex) to maintain their competitive moats as raw intelligence risks becoming a commodity. • Extending the availability of flagship models (like Anthropic's Fable) to stay competitive against surging open alternatives.

Takeaways

• Potential revenue and margin headwinds for closed frontier labs if open-weight models continue to close the capability gap. • Investors should monitor how effectively frontier labs can monetize their proprietary software wrappers and harnesses (such as coding assistants) rather than just raw model capabilities.