Sundar Pichai, the CEO of Google, recently took a reporter on a tour of the Googleplex, the tech giant’s headquarters in Mountain View California. The tour provided a glimpse into the intricate world of artificial intelligence (AI) development, showcasing both the advancements and challenges faced by the industry. At the heart of this innovation are Google’s Tensor Processing Units (TPUs), super specialized chips that are a purpose-built machine behind AI. The debate over these chips brings into sharp relief the unprecedented scramble among technology companies to obtain chips that are vital silicon components.
Pichai showed the reporter around the shiny new campus. He illustrated the important role that TPUs have in influencing the future of AI. There have been several versions of these chips, with the most recent update being the Ironwood version. According to Pichai, Google’s TPU cluster is hard at work creating that chip. This new chip could be the one that powers all of Google’s generative AI functionalities in search. The laboratory specifically for testing these TPUs is the size of a five-a-side football pitch. It hums with a dense lattice-work of rainbow-colored wires and a tangle of pulsating deep blue lights.
The race for AI supremacy isn’t only being fought at Google. Other tech giants, such as OpenAI, are competing for their bite of this golden goose market. Sam Altman, the CEO of AI giant OpenAI, recently recommended that governments should go so far as to build and own the AI infrastructure themselves. He expressed concerns about the current landscape of AI technology, stating, “I do not think the government should be writing insurance policies for AI companies.” Rather, he suggests that AI infrastructure be owned by and developed through the government to help produce more equitable and responsible results.
The stakes have never been higher. Combined, Google and the other four tech giants have an incredible value of $15 trillion. This unprecedented concentration of wealth and power has alarmed even some financial institutions. The International Monetary Fund (IMF) highlighted that the market value of tech AI firms is more concentrated than during the dot-com bubble in 1999. Given these high valuations, fears are growing that we are due for a painful market correction. Even the Bank of England is cautioning against a possible “sudden correction” in international financial capital.
In that discussion, Pichai recognized the volatility that’s inescapable in these investment cycles. “It’s true when we go through these investment cycles, there are moments we overshoot collectively as an industry,” he stated. This speaks to a much larger, growing fear among leaders in the industry about how speculative today’s AI investments and promises are.
The unprecedented energy demand to power and cool these data centers is equally alarming. Projections show that by 2030, data centers globally will use the same amount of electricity India used in 2023. This rapid, exponential growth makes clear the urgency for more sustainable practices to thrive within the technology sector.
Even considering these challenges, Pichai expresses optimism about the potential of AI. He described AI as “the most profound technology humanity has ever worked on,” indicating his belief in its transformative power. He warned against thinking that we can just put AI technology in place and ignore the need to enrich the information ecosystem around it. “If you only construct systems standalone and you only rely on that, that would be true,” he remarked. Which is why I think we have to make the information ecosystem we’re creating to be much, much richer than having AI technology making that the only product in there.
At a recent dinner at Nobu in Palo Alto, Larry Ellison, founder and head of Oracle, described his and Elon Musk’s efforts to persuade Nvidia’s Jensen Huang to sell them more top-performing chips. Ellison humorously recounted their approach: “I would describe the dinner as me and Elon begging Jensen for GPUs. Please don’t do that – please take our money – no, no take even more. You’re not taking enough. We need you to take more, please!”
As companies scramble for resources and navigate the complexities of AI development, they must contend with market dynamics and regulatory considerations. It’s hard to say what the future landscape of AI will look like. Yet, as much as anyone, those who work on the frontline are acutely attuned to the promise and peril present in this fast-moving new landscape.