Amazon (AMZN) and Alphabet’s Google (GOOGL) are entangled in a fierce battle for a lucrative contract to develop a data infrastructure in Israel. The $266 million contract is for developing a supercomputer to be used to train large artificial intelligence models. The development of the supercomputer is part of Israel’s bid to be a global leader in the fast-evolving AI sphere.
Israel Supercomputer
Supercomputers are next-generation devices that can process exponentially more data than traditional computers and can crunch numbers at lightning-fast speeds. In order to run intricate simulations, answer questions about nature and climate change, and create new AI-driven technologies, they extensively use graphics processing units (GPUs), which provide high processing and computational capabilities beyond the capabilities of standard computers.
Amazon and Google are familiar with lucrative tenders in Israel’s robust tech industry. In 2021, the government selected two companies to build and develop cloud-based regional data centers for Project Nimbus. While Israel is home to over 9,000 startups, of which over 2,200 are AI-based firms, it is looking to enhance its edge around revolutionary technology as it looks to boost its education, defense and healthcare sectors.
AT&T Power Warning
Meanwhile, AT&T (T) CEO John Stankey warned that the US is not ready for the energy consumption the AI revolution will fuel. According to the CEO, the AI revolution is increasingly straining current energy sources and promises to be the next big social and economic issue. The CEO fears the country has enough power to make all the work around artificial intelligence work.
The warning comes amid a report by Gartner that affirms AI energy demand is well poised to exceed current power utilities in the US by 2027. The study also showed that up to 40% of data centers could face power shortages in the next three years. According to a previous study by Goldman Sachs, data centers will need 160% more electricity by 2030.
The report cautioned that the United States hasn’t experienced such a surge in power demand since the early years of this century. For example, the investment bank claims that a single ChatGPT query consumes ten times as much power as a Google search. The second problem with AI’s extreme resource consumption is that it may undermine global and corporate leaders’ pledges to achieve net zero and combat climate change.
Microsoft’s Facebook Poaching
Separately, Microsoft has moved to address its inefficiency in bringing online data centers faster to address its artificial intelligence needs by hiring an engineering chief who kept Facebook’s infrastructure running smoothly. Jay Parikh will become a member of the senior leadership group and answer to Microsoft CEO Satya Nadella.
After joining Facebook in 2009, Parikh worked on data center and technical infrastructure projects for over ten years, helping the company expand into the largest social network in the world. During Parikh’s tenure, Meta began constructing over a dozen data centers worldwide. Meta remained one of the few businesses able to construct state-of-the-art server farms on a large scale. At the same time, many tech companies disconnected their own data centers to rent computing power from Amazon.com Inc. or Microsoft (MSFT).
