| WEEKLY ROUNDUP Everyone Chased GPU Stocks, but Now AI Is Turning Elsewhere VIEW IN BROWSER Hello, Reader. Many products or pieces of hardware usually age quickly in the tech world… but that doesn’t mean they can’t find new meaning. At least that’s the case with central processing units (CPUs). CPUs used to be the most important component in managing a computer’s activity, speed, and capabilities, acting as the “brain.” Their importance remained, but when artificial intelligence emerged, tech companies quickly learned that graphics processing units (GPUs) were better suited to run AI models, performing technical calculations faster and more efficiently than CPUs. That made Nvidia Corp. (NVDA) the AI chip king… and its investors hugely successful. Over time, demand for GPUs rose, creating a compute shortage that we discussed last Thursday in Smart Money. That put Nvidia in a good spot, to say the least. But now, there’s a fundamental change in compute needs thanks to the rise of agentic AI. Because AI agents are task-oriented, CPUs are the ideal fit for running them, as they have fewer powerful cores than GPUs when running consecutive general-purpose tasks. And Nvidia is once again ready to profit from this shift. It’s already sending that message by recently boasting about the success of its deployment of Grace CPUs in powering Meta Platforms Inc.’s (META) data centers, a partnership that began just last month. In a press release, Nvidia stated the CPUs’ ability to improve performance per watt in Meta’s data centers. Today is the first day of the Nvidia GTC AI conference, an expo where thousands of developers, researchers, and business leaders share and discuss tech’s latest AI breakthroughs. Leading up to the event, Nvidia’s head of AI infrastructure expressed the “exciting opportunity” CPUs present, and noted that they are “becoming the bottleneck in terms of growing out this AI and agentic workflow.” This afternoon, the company revealed its new 88-core Vera data center CPUs, which deliver 50% performance gains over traditional CPUs. Nvidia also announced details about its new Vera CPU Rack, which integrates up to 256 liquid-cooled CPUs into a single rack for CPU-centric workloads, improving core sustainability and enabling twice the energy efficiency. Now, many investors will join Nvidia as it profits from the rise of agentic AI… but I’d like to offer you an even more profitable approach. It’s true that Nvidia is a spectacular, industry-leading company. But because it is trading at a spectacularly high valuation, investors are putting themselves in a vulnerable position. Nvidia may be the most obvious investment choice to ride the CPU demand wave, but that doesn’t mean it will be the most profitable. But before I share how investors should prepare for agentic AI’s takeover, let’s look at what we covered here at Smart Money this past week… | 
Post a Comment
Post a Comment