EVOBITS IT
Keeping AI cool The technical requirements for AI computing have also prompted innovations in cooling system design . For EvoBits IT , its experience in cryptocurrency mining operations has proved valuable in this task , as the company is successful in managing highdensity computing environments .
“ We discovered that traditional data centre design standards don ’ t fully apply to AI systems . Typically , you ’ d calculate cooling based on thermal units that need dissipation , with a standard ratio of airflow per thermal unit ,” Silviu explains . “ With AI systems , the required airflow is significantly higher than what thermal dissipation calculations would suggest .”
The company ’ s current facility implements strict airflow management protocols , including aisle containment systems that separate hot and cold air paths . These systems are informed plans for the new facility , having shaped the design of smaller server clusters with both hot aisle containment and raised floor cold aisle containment – systems that direct cool air to equipment and remove heated air efficiently .
EvoBits IT also employs indirect evaporative cooling , a system that uses water evaporation to remove heat without mixing outside air with the data centre ’ s internal environment . This is able to achieve a Power Usage Effectiveness ( PUE ) ratio below 1.15 , meaning overhead energy consumption adds less than 15 % to computing power consumption .
The company has proposed a modular 18MW facility which would be developed
sustainabilitymag . com 95