Artificial Intelligence · 8 July 2025

AMD makes energy efficiency its weapon against Nvidia

Data centres now consume almost as much electricity as all of Germany. Chip company AMD is turning that into a sales pitch – and trying to close the gap on market leader Nvidia with power-saving silicon.

🔌 Data centres now consume almost as much electricity as all of Germany. Chip company AMD wants to turn that into an advantage – pitching new, particularly power-efficient silicon.

🤖 In the global race for the best chips for artificial intelligence, Nvidia and CEO Jensen Huang are clearly ahead. The only seriously credible rival is AMD. But the company under Lisa Su’s leadership is having a hard time standing up to Nvidia’s dominance.

⚡ Now AMD is leaning on a new argument: energy efficiency.

While US tech giants like Google or Meta are turning to nuclear power plants to feed AI’s huge appetite for electricity, AMD is trying to score with technological progress.

🛠️ In a conversation with AMD’s chief technology officer Mark Papermaster, he listed a number of tricks AMD has used to lift energy efficiency.

🧠 Together with our chip expert Joachim Hofer I took a closer look at the technical innovations. The full piece is in Handelsblatt.

Published in Handelsblatt

July 5, 2025 ·Artificial intelligence

Artificial intelligence – these are the chips with which AMD wants to attack market leader Nvidia

Lower power consumption, lower cost: chip maker AMD is raising big expectations for its new chip generation. They have the potential to change the entire AI market.

Machine translation. The verbatim German original is on handelsblatt.com (link below).

The annual electricity consumption of the AI servers of Amazon, Microsoft and Google is almost as high as that of the whole of Germany. And the gap is closing fast, according to the International Energy Agency. Data centres globally currently consume 415 terawatt-hours (TWh) per year; Germany consumes 474 TWh.

And the International Atomic Energy Agency estimates in a recent study that the power consumption of AI data centres will more than double again by 2030. For Gartner analyst George Brocklehurst the conclusion is clear: "Energy is by far the most important factor when we talk about new AI factories."

Energy costs a lot of money, especially in Germany. And generating the required electricity can accelerate climate change because most of it still comes from fossil sources. But Mark Papermaster sees an opportunity – one that starts with energy consumption itself. "The challenge is solvable," the chief technology officer of chip company AMD told Handelsblatt. And that could have consequences for the entire market.

Nvidia currently dominates the global business in AI chips. With ever-more-powerful processors, the world's most valuable company equips the data centres. To win market share, AMD is now leaning hard on efficiency. Papermaster sees his technology as a key selling point in catching up with Nvidia.

AMD CEO Lisa Su estimates the market for AI chips in 2028 at a revenue of more than $500 billion. For comparison: last year, the entire chip industry generated revenues of around $630 billion.