Apple reportedly developing AI chips
Apple now reportedly developing its own chip to run AI software inside data centers. According to the Wall Street Journal, Drew Dubose is taking a look at the AI chip arms race for today’s tech Check. Morning Dee. Good morning, Carl. I know you’re going to like this name. It’s apparently codenamed Project ACDC for Apple chips in data center, and here may lie Apple’s advantage in the Gen. AI race. This company has the most experience and success in designing its own custom silicon, even if it has been a laggard in terms of laying out that broader AI strategy to investors. At the Apple event in Cupertino this morning that Steve was covering, the company debuted its latest M Series processor, the M4, which it called a, quote outrageously powerful chip for AI. But AI chips for data centers? That’s a different beast and an unproven area. And while that M line of silicon has improved power efficiency performance across Apple devices, Apple has struggled in other areas like developing its own cellular wireless chip to replace Qualcomm silicon. Now the Journal reports that Apple is working with Taiwan Semi on a server chip that runs AI models. So that’s the inference part of AI versus training. But if that is successful, that could give Apple more control over its AI products and ambitions when it does lay that out expected at WWDC in a few weeks. Now, as we’ve discussed, nearly all the mega caps are trying to develop their own Gen. AI custom in house chips. It’s about cost and efficiency and countering Nvidia’s dominance. Billionaire investor Stanley Druckenmiller talked about Gen. AI and trimming his NVIDIA position on Squawk this morning. We’ve had a hell of a run. A lot of what we recognized has become recognized by the marketplace. Now long term, we’re as bullish on IAI as we’ve ever been. The big payoff might be four to five years from now. So AI might be a little overhyped now, but underhyped long term, so he just trimmed his position. He said he wouldn’t be surprised if he was holding on to NVIDIA for the next 5 to 10 years in some way. But to that sentiment, that longer term sentiment. Of course, nobody’s close to replacing NVIDIA GPU’s anytime soon, but Apple and other mega caps, other chip makers too. They’re playing the long game here. Silicon is a very key piece of that strategy, and that willpower their apps and products, which is ultimately what they’re looking to make. Girl, a couple of reflections. One, I was fascinated by Druckenmillers answer that he can’t imagine not owning NVIDIA on or off for the next 5 to 10 years. And then just a reminder that Apples only been in the chip business for what, a decade Maybe only, yeah. But I mean, it’s a bit of a head start over some of the mega caps. This is true. But I mean, you look at the success that it’s had with that M series, right, and how it’s made improvements across iPhones and iPads and even the laptops. Apple would tell you that it’s been working on AI chips for a long time as well. But this is where it’s been a lot harder, right? We talked to a lot of the different mega caps about the chips they’re making, Google’s TP us in house, and they say that they’re getting better and better. But I think the real test is who’s using them, right? They’re all being used inside of their cloud infrastructure. Can you use it for inference? Can you use it to build the foundational models? The race is happening at this level and that is going to set the ground for the race. When we get into the apps and the products, which everyone’s sort of looking, looking forward to, it’s amazing. I mean, back then it was just going vertical to get a better hold of their own supply chain for goods. Yeah. Now it’s a much more, much bigger Tam. Overall, a fascinating, A fascinating look at an important story today. D Thanks, Did Ribosa.