Mystery buyer places $300 million order for diamond-cooled AMD GPUs to tackle AI data center heat

2 Sources

Share

An undisclosed US buyer has placed a $300 million order for AI servers featuring AMD Instinct MI350X GPUs cooled with lab-grown diamonds. Akash Systems, backed by Peter Thiel, claims the innovative cooling solution removes heat five times faster than copper and delivers up to 22% higher performance per watt, addressing the thermal challenges of high-performance AI hardware.

Undisclosed US Buyer Commits $300 Million to Diamond-Cooled AMD GPUs

An undisclosed US buyer, likely a major tech company or data center operator, has placed a $300 million order for AI servers built around AMD Instinct MI350X GPUs featuring an innovative cooling solution that uses lab-grown diamonds

1

2

. The hardware comes from Akash Systems, a Peter Thiel-backed startup that has developed what it calls Diamond Cooling technology, with manufacturing handled by Taiwan-based MiTAC Computing. The substantial investment signals growing industry interest in addressing the thermal challenges of high-performance AI hardware as AI data centers struggle with escalating power demands and heat management issues.

How Diamond Cooling Tackles Heat Management in AI Servers

The diamond component is integrated directly into the server between the GPU and the heat sink, acting as a new layer in the data center cooling stack

1

. Diamonds possess the highest thermal conductivity of any known material, removing heat five times faster than copper, the industry standard material for heat management

1

. This synthetic diamond material spreads heat away from hotspots more efficiently than conventional solutions, helping to enhance heat dissipation across the entire system. Akash Systems claims the technology enables AI servers to run throttle-free at higher ambient temperatures while helping to reduce power consumption

1

. The company reports that diamond cooling can maintain high performance at around 120°F (48.8°C) ambient temperature and potentially operate with up to 100% less power dedicated to cooling thanks to lower GPU temperatures

2

.

Source: TechSpot

Source: TechSpot

Performance Gains and Energy Efficiency Benefits

Akash Systems claims the diamond-cooled AMD GPUs deliver up to 22% higher FLOPs per watt and up to 15% throughput improvement compared to traditional cooling methods

2

. The servers pair AMD Instinct MI350X GPUs with dual 5th-gen AMD Epyc 9005 CPUs, AMD Pensando Pollara 400 AI NICs, and the latest ROCm stack

2

. Travis Karr, AMD's Corporate Vice President for Commercial and Enterprise AI, stated that the partnership with MiTAC's global deployment capabilities enables data centers to realize impactful compute density and breakthrough energy efficiency

1

. The company has previously deployed its technology in satellite radios, making them 3x smaller and consuming 60% less power

1

.

From Nvidia to AMD: Akash Systems Expands Diamond Cooling Deployments

This isn't Akash Systems' first major deployment. Last month, the company announced the first shipment of diamond-cooled Nvidia H200 GPU servers to NxtGen, India's largest sovereign cloud provider, which are already operational

1

2

. The company claims to have established a robust supply chain for customized lab-grown diamonds at a very different price point from mined diamonds, making the technology economically viable

1

. Theoretically, the energy cost savings would offset any potential additional cost of the diamond cooling technology. The $300 million order represents a significant validation of the approach, though independent testing will be needed to verify the performance claims. As AI workloads continue to push the limits of existing cooling infrastructure, alternative materials like lab-grown diamonds may become standard in next-generation data center designs, particularly for operators seeking to maximize compute density while managing electricity costs and environmental impact.

Today's Top Stories

TheOutpost.ai

Your Daily Dose of Curated AI News

Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Spend less time searching for the latest in AI and get straight to action.

© 2026 Triveous Technologies Private Limited
Instagram logo
LinkedIn logo