Emerald AI Collaborates with Google and Nvidia to Revolutionize Data Center Integration with the Grid
Emerald AI, a burgeoning player in grid software, has established a partnership with tech giants Google and Nvidia, leveraging the startup’s innovative technology to allocate more space on the grid for data centers.
The newly formed AI Energy Management Alliance (AEMA) aspires to embed demand response — a strategy whereby data centers temporarily curtail their electricity consumption — into the fabric of data center development.
By suspending non-essential operations and redistributing some computational tasks, AEMA contends that it could enable the integration of an additional 100 gigawatts of data centers into the grid.
Joining the initial triad, Anthropic stands alongside various utility entities, including AES, Constellation, National Grid, and NRG Energy.
Demand response is a time-honored principle that utilities have harnessed for decades. Its efficacy stems from the grid’s architecture, designed primarily for peak load scenarios. Throughout much of the year, actual demand significantly lags behind the grid’s maximum capacity.
Traditionally, this demand response mechanism engaged substantial electricity consumers, such as factories, which during peak demand periods would consent to decrease their electricity usage, typically through temporary production halts or by switching to backup generators. In recompense, utilities compensated these users generously for their cooperation.
Modern data centers also partake in such programs, frequently activating backup generators during peak times.
Emerald AI is positing a cleaner alternative to conventional diesel generators. The company’s software orchestrates requests from utilities with data centers, enabling the latter to pause non-critical tasks or reroute loads to other data centers operating within available grid capacity.
Emerald AI is not alone in this endeavor; Google has been developing similar solutions, while Enel X allows data centers to utilize their uninterruptible power supplies to alleviate peak power demands.
Goldman Sachs, in a study published last year, indicated that restricting grid usage to 90% for limited intervals could free up 76 gigawatts of capacity for data centers.
However, Emerald AI possesses a distinctive advantage. Its software facilitates a direct link between utilities and data centers, allowing for a prompt response to requests, akin to the functionality of batteries.
Furthermore, the startup has recently secured $150 million in a Series A funding round led by Energize Capital and DCVC, providing the necessary financial backing to expand its technological reach.
Given the erratic nature of AI computational workloads and data centers’ capability to swiftly adjust their power consumption, the establishment of AEMA is both timely and necessary.
This coalition also aims to assist tech firms and utilities in identifying new locations for data centers, a complex undertaking for both parties.
While this advanced approach to demand response holds significant potential, it is important to note that it will not entirely rectify the current challenges facing the grid.
Ayse Coskun, Emerald AI’s chief scientist, acknowledged to TechCrunch that while the technology offers a means to mitigate the industry’s reliance on new power generation sources, it cannot eliminate this necessity entirely.
Source link: Techcrunch.com.
