Today in AI: Infrastructure & Enterprise Adoption

June 22, 2026

The AI infrastructure race heats up with a major efficiency breakthrough from NVIDIA, while OpenAI pulls off a massive enterprise coup. Meanwhile, the community debates the open vs. closed model landscape.

• NVIDIA reveals servers with 45°C liquid cooling

NVIDIA's new AI servers can run cooling liquid at a scorching 45°C, a significant increase over typical systems. This isn't just a tech spec—it dramatically reduces the energy and cost of cooling massive data centers, a critical bottleneck for scaling AI infrastructure. (NVIDIA)

💰 Samsung deploys ChatGPT Enterprise worldwide

Samsung Electronics is rolling out ChatGPT Enterprise and Codex to its global workforce. This is a massive validation for OpenAI's enterprise strategy and shows how AI is moving from experimentation to core operational tools at the world's largest companies. (OpenAI)

• Hacker News debates 'minimal downside' to open models

A discussion on Hacker News argues there is 'minimal downside' to switching from closed to open-source AI models for many use cases. This sentiment reflects the growing confidence in the open-source ecosystem and its ability to match proprietary models for specific tasks. (Hacker News)

🧠 Apertus pitched as an open foundation for 'Sovereign AI'

The Apertus project is being discussed as an open foundation model designed to enable 'Sovereign AI'. This taps into the rising global trend of nations and organizations wanting AI systems they can control, free from dependence on a few corporate giants. (Hacker News)

The takeaway: **NVIDIA's high-temperature cooling breakthrough is a pivotal engineering win that will lower the cost and energy footprint of training the next generation of massive AI models.**