Chinese supercooling tech could boost AI race

Chinese researchers have developed a liquid cooling system using ammonium thiocyanate that achieves rapid cooling in 20 seconds. By mimicking the squeeze of a wet sponge through pressure changes, it instantly absorbs massive heat, potentially offsetting the soaring energy demands of AI data centers.

Researchers at the Institute of Metal Research, part of the Chinese Academy of Sciences, led by Li Bing, have harnessed the unique behavior of ammonium thiocyanate in water under pressure to create a liquid cooling system. It mimics squeezing a “wet sponge”—releasing pressure triggers rapid redissolution of the salt, absorbing massive amounts of heat almost instantly.

In experiments, a saturated solution cooled by 30 degrees Celsius (54 Fahrenheit) within seconds at room temperature, and in hotter environments, the drop exceeded 50 degrees. This technology could power critical infrastructure, offsetting the soaring energy consumption and cooling demands of AI data centers.

Amid the US-China AI race, with figures like OpenAI's Sam Altman driving US efforts, this innovation from China, published in Nature, highlights potential edges in fluorocarbon gases and supercooling for data centers.

関連記事

Chinese scientists have developed a supercooling innovation that boosts the performance of gallium nitride chips used in military radar by 40%. This technology, from Xidian University, enhances radar detection in stealth aircraft without increasing chip size. It also offers wider signal coverage and lower power costs for mobile networks.

AIによるレポート

研究者らが10億分の1秒で光パルスを発射する超高速レーザー技術を開発し、1,000倍強い構造を1,000倍速く作成可能にした。この革新的手法は、フォノン散乱距離を制御してチップの熱伝導率を標的とし、高性能コンピューティング、量子デバイス、AIチップ冷却への応用を提供する。ファンや液体冷却に頼らずチップの熱処理方法を変革する。

Rising AI demand is fueling global data center growth, with significant implications for power and sustainability. In the Philippines, the government is pushing for more data centers to achieve digital transformation goals, but the country's hot climate poses challenges for cooling and energy use.

AIによるレポート

オピニオン記事で、ネイサン・ロード氏は、米国が信頼できるエネルギー源、特に天然ガスを確保しなければ人工知能でリードできないと主張。中国の優れた電力生産を強調し、シェール・クレセントなどの地域で燃料供給源近くにデータセンターを建設するよう促す。このような対策なしに、米国はグローバルAI競争で後れを取るリスクがある。

 

 

 

このウェブサイトはCookieを使用します

サイトを改善するための分析にCookieを使用します。詳細については、プライバシーポリシーをお読みください。
拒否