Welcome to the Eye on AI newsletter! In this issue, we cover Ilya Sutskever’s new role as CEO of Safe Superintelligence, the impact of Chinese AI companies on U.S. dominance, the heated AI talent bidding war at Meta, and Microsoft’s full commitment to AI in its sales overhaul.
As a scorching early-summer heat wave enveloped New Jersey last week, a concerning prediction from Accenture caught my attention: AI data centers are projected to increase their carbon emissions by 11-fold by 2030. The report suggests that over the next five years, AI data centers could consume an estimated 612 terawatt-hours of electricity, equal to Canada’s annual power usage, leading to a 3.4% rise in global carbon emissions.
The strain extends beyond energy consumption to water usage, with AI data centers projected to consume over 3 billion cubic meters of water annually, surpassing the freshwater withdrawals of entire countries like Norway and Sweden. While the report, “Powering Sustainable AI,” offers suggestions to mitigate these issues, the relentless expansion of Big Tech’s AI data centers worldwide raises doubts about the industry’s commitment to sustainability.
Recent developments, such as OpenAI’s partnership with Oracle for computing power and Meta’s plans to raise $29 billion for new AI data centers in the U.S., underscore the escalating competition for AI dominance. Meta CEO Mark Zuckerberg emphasizes the need for rapid expansion of AI data centers to match China’s aggressive efforts, warning that America risks losing its AI edge without significant infrastructure growth.
The urgency surrounding AI development is echoed by the U.S. government, with White House AI and Crypto Czar David Sacks advocating for streamlined permitting and increased power generation to support AI infrastructure. Sacks believes that prioritizing energy and data center expansion is crucial for America to maintain its competitive position in the global AI race.
The Biden administration has made it extremely difficult, according to him. He believes the U.S. must quickly develop faster technologies. Accenture, on the other hand, is advising its clients to expand and design their AI data centers responsibly to balance growth with environmental concerns. They have introduced a new measure, called the Sustainable AI Quotient (SAIQ), to assess the real costs of AI in terms of investments, energy consumption, COâ‚‚ emissions, and water usage. This metric aims to help organizations understand the value they are getting from their AI investments and track their progress over time.
I had a conversation with Matthew Robinson, managing director of Accenture Research and co-author of the report, who expressed hope that their predictions would not come true. He explained that they projected future energy consumption based on expected AI chip installations and data center energy needs. Regional electricity generation, energy sources, and emissions data were also considered, along with water consumption per unit of electricity generated by AI data centers. Robinson stressed the importance of discussing strategies to avoid the predicted outcomes.
Accenture recommends improving the energy efficiency of AI workloads and data centers through various means, including using low-carbon energy sources and innovative cooling techniques. They suggest using AI judiciously by opting for smaller models and implementing pricing models that incentivize efficiency. Additionally, they advocate for better governance over AI sustainability initiatives.
It might be challenging for major players in the AI industry, such as Big Tech companies and well-funded startups, to slow down their pace to address these concerns. However, it is not impossible. For instance, Google recently disclosed in its sustainability report that its data centers are consuming more energy than ever before. Despite this, Google has reported progress in reducing data center energy emissions and improving efficiency. The company has significantly increased its compute efficiency and is approaching optimal data center efficiency levels.
Robinson remains cautiously optimistic about the future.
During a visit to IBM’s Yorktown Heights campus in New York, I explored a midcentury modern architectural wonder crafted by the same designer behind the St. Louis Arch and the iconic TWA Flight Center at JFK Airport. This visit was in connection with my coverage of IBM’s recent resurgence for this year’s Fortune 500 issue. In my article, I described the experience of walking through the IBM research center as a journey through two distinct worlds. The building features the elegant steel and glass curves of Saarinen’s design, complemented by massive walls constructed from stones gathered from the nearby fields, and original Eames chairs scattered throughout discussion areas. The juxtaposition of 20th-century modernism and cutting-edge technology is striking, particularly in the presence of IBM’s state-of-the-art quantum computer, one of the most advanced globally, located in the collaboration area and operational lab. The quantum computer emits a steady hum from its cooling system, resembling a sleek, massive refrigerator. For more AI news, please refer to the contact information below. Sharon Goldman, sharon.goldman@fortune.com, @sharongoldman. This article was originally published on Fortune.com.
