This will be music to the ears of Meta investors, who’ve already seen the company sink $140 billion into its artificial intelligence projects.
Today, Meta has announced that it’s breaking ground on a new 1 gigawatt data center project in Lebanon, Indiana, which Meta says “represents an investment of over $10 billion in data center infrastructure and the surrounding community,” one of its largest infrastructure investments to date.
The project is Meta’s 30th data center initiative in the U.S., with several launching over the past year, as the company continues to push towards the next stage of AI development.
As explained by Meta: “This new data center is our second site in Indiana, and is designed to deliver 1GW of capacity once operational. As AI advances and compute demands continue to grow, gigawatt sites like this one will be critical to advancing the technology that supports our core business as well as our AI ambitions. Building at this scale creates the flexibility to support both goals while enabling technology with higher bandwidth, lower latency, and improved reliability.”
So, cool, another few billion to power Mark Zuckerberg’s AI vision, which he believes will lead to a new future of digital connectivity and productivity, in a range of ways.
And given the amount that Meta has committed to the concept, it better, because Meta is on track to spend hundreds of billions on AI infrastructure over the next few years, and that’s before you consider ongoing maintenance and development, which will continue to keep those costs high.
Which means that Meta needs to start making money out of its AI projects, and soon, otherwise it could become an albatross, and an expensive pipe dream, which fails to deliver on the promises of the modern Silicon Valley prophets.
The longer-term goal of AI, of course, is to create artificial general intelligence (AGI), or machines that can actually think like a human, and with far expanded digital capacity, these systems could quickly surpass human development, which could eventually lead to new breakthroughs that it would have taken decades for our mushy brains to come up with.
That’s Zuckerberg’s long-term vision, and has been for a long time, with Meta developing AI for years before OpenAI brought machine learning into public consciousness, and made that what we see as AI.
Which it isn’t, and that’s an important distinction to make. Tools like ChatGPT are not thinking, they don’t have any concept of the answers that they provide, it’s just advanced pattern matching, enabled by improved understanding of human language.
Which is helpful in some contexts. However, the risk of the current wave of AI tools is that they give the perception of thought, without actually thinking, which means that their outputs are risky, because it’s interpreting whatever data inputs it has, which can be massively flawed, and utterly incorrect, at times.
True AI will extend beyond this to give us more reliable, more valuable insight, which could eventually change the world. But we’re not there yet, and we might never be. And we’ll definitely never reach that next threshold without significant investment.
Which is why Meta is spending so much on AI data center projects, and pushing towards the next stage. That also means that the current AI tools we have are something of a distraction, but Meta also needs people to use these tools so it can start making money from its investments.
Will Zuckerberg’s big spending eventually pay off? I don’t think that anybody knows for sure, but clearly, Zuck and his crew think that this is possible, and they’re spending more than the GDP of most nations to reach that goal.
Which will no doubt make some Meta investors a little uneasy. But this is where we’re at, and if it’s an AI bubble, and things don’t pan out as expected, it could be an expensive back-end for AI image generators.
