The AI giants of the tech world are apparently standing around with their checkbooks out, waiting for Nvidia to deliver its next-gen architecture, Blackwell. CEO Jensen Huang first unveiled Blackwell way back in March and said it would be shipping later this year. In a new interview, Jensen says production has begun, and demand has been "insane" amongst the top-tier cloud companies such as Microsoft and Meta.
Jensen's comments were made in a new interview with CNBC, where he discussed the company's current status as a darling of Wall Street. In the short video, flagged by Wccftech, he states Blackwell production has begun, and demand has been "insane" because everyone wants "the most" and "[wants] to be first," which is no surprise. Given the current AI arms race, it seems like simple math that if you have the newest hardware from Nvidia, you can scale your AI operation quicker than if you were using an older architecture like Ampere.
That's the thought process that Jensen is promoting with Blackwell too, and it's not a new thing. Back in 2018 he was famously roasted for saying, "The more GPUs you buy, the more money you save." With Blackwell, Jensen is taking this concept even further and drawing a straight line from its increase in throughput compared with the previous architecture named Hopper to a company's potentially increased earnings by deploying its technology.
He states that Blackwell offers 2-3x the throughput of Hopper, so companies who buy Blackwell can also expect their throughput and revenue to increase by the same amount, which seems ambitious. It's so early in the AI revolution, for lack of a better word, that nobody knows yet just how profitable investments in AI will be over the long haul. Despite AI's murky prospects, every big tech company is building new data centers or upgrading existing ones to train large language models to get in on the action.
We'll start to find the answer to this question in 2025, though, as AI is expected to be ubiquitous. It'll be in all of the newer smartphones and spread across Copilot+ PCs from Intel, AMD, and Qualcomm. We're sure there are other "touchpoints" that will arise as well that we haven't contemplated yet, as companies of all sizes are seeking to add AI to their products to capitalize on the buzz that ChatGPT has generated.