The AI ​​revolution on PCs and smartphones is canceled for now – sky-high sales have not happened

Micron was unable to achieve the expected results for the last quarter and was forced to lower its forecast for the current quarter – authoritative analyst Daniel Newman believes that this is not a problem of just one manufacturer, but of the entire technology industry: the revolution of PCs and smartphones with artificial intelligence did not happen, and there is no need to wait for it yet.

Image source: microsoft.com

Much of Micron’s problems were caused by weaker-than-expected markets for memory components for PCs and smartphones. Micron’s revenue for the quarter amounted to $8.709 billion versus analysts’ expectations of $8.721 billion; in the current quarter, the company expects to earn $7.9 billion against the $8.98 billion predicted by Wall Street analysts – due to such a strong discrepancy, the manufacturer’s shares fell by more than 16%. These indicators should be paid attention to, but they do not foreshadow tragedy yet, Mr. Newman believes – this is not “the beginning of the end for the AI ​​industry” and not the collapse of Nvidia.

Micron relies heavily on the HBM memory market, which is expected to grow to $16 billion this year and reach $100 billion by 2030, but its main source of revenue remains making memory chips for PCs and smartphones. “However, the core business is shrinking as PC and smartphone shipments lag [forecasts] and Micron must cope with customer inventories that are selling slowly, which will lead to even lower orders/sales in this and subsequent quarters. The bad news is that the supercycle of AI PCs and AI smartphones has more or less failed,” Newman writes.

In 2023 and 2024 new AI features will trigger high demand for PCs with their support, but this did not happen. Demand for AI-powered PCs is driven not by AI, but by faster CPUs and GPUs, a September report from IDC Research found. The need to replace Windows 10 PCs with Windows 11 models in the new year will have a stronger impact on PC sales growth than AI, Trendforce believes. Qualcomm is clearly having trouble with its new Snapdragon X chips for Copilot Plus laptops: in the third quarter, the company managed to capture only 0.8% of the PC market, selling 720,000 units. Acting Intel CEO Michelle Johnston Holthaus said that the return rate for laptops powered by Snapdragon X is too high, but Qualcomm did not agree with her assessment.

Today, the manufacturer of Arm processors expects the release of new models on Snapdragon X, which, with the same performance of the AI ​​accelerator, will offer prices of $700 – the models on sale today have price tags of $1000, and a reduction of $300 seems significant. If demand for current AI PCs was strong enough, it wouldn’t make sense for Qualcomm to switch to a lower price range – it makes sense when demand is weak in the higher price segment. PC AI has a major problem: running AI locally on a PC isn’t much use these days. Existing AI software is more of a hobbyist endeavor, with popular services like ChatGPT running in the cloud and running without AI accelerators.

A PC with AI is quickly becoming just a PC because AI accelerators are becoming an industry standard, and AI has never become a selling point for new products – in the same way that multi-core processors, integrated graphics, and SSDs have become the standard for PCs. Consumers have no incentive to specifically buy PCs with AI, but they won’t buy PCs without AI either, because this will imply an outdated processor. There are no signs of a collapse in the AI ​​industry either: demand for HBM memory remains high, it sells well, but remains in demand only in server products from Nvidia, Broadcom, AMD and Marvell. The lack of excitement among consumers when running AI locally on PCs and smartphones certainly doesn’t look good on the tech industry, but it doesn’t point to a tragedy either—low demand for HBM would be much worse. Although if the AI ​​industry ends up being a bubble and bursts, the first harbinger of the collapse will become obvious – the lack of explosive consumer demand for local AI startups.