In the world of hardware manufacturers, net profit rarely tells the whole story. Lenovo ‘s third quarter financial report on Thursday is proof of this. Although net profit attributable to shareholders fell 21% to $546 million, investors got the signal they were waiting for: the Chinese giant is effectively shifting its centre of gravity towards artificial intelligence before the traditional PC market hits the wall.
The company’s financial results expose a classic transformation dilemma. Revenues grew by 18% year-on-year, reaching $22.2 billion and significantly beating analyst forecasts. However, this success comes at a price. Lenovo is facing a growing shortage of memory chips, which has forced the company to raise prices on end devices. CEO Yang Yuanqing outright admits that PC sales volumes will come under pressure, and that a flight into the AI infrastructure segment is expected to rescue profitability.
The most interesting dynamics are taking place within the server division. The digital infrastructure group reported a 31 per cent increase in revenue, driven by deployments of solutions based on the Nvidia GB200 architecture. Although the segment generated an $11 million operating loss, this is a cost that Lenovo is consciously making part of its strategy to build scale. The company assumes that the AI infrastructure market will triple by 2028, and the real money no longer lies in training models, but in deploying them, or so-called inference.
HR and operational moves confirm this turnaround. Lenovo has earmarked $285 million for restructuring, with a target of $200 million in annual savings. These funds will be redirected to the development of a portfolio of enterprise servers, created in partnership with AMD, among others.
The lesson from Lenovo’s report is clear. Even dominance of the PC market, which still accounts for 70% of the group’s revenue, does not guarantee security in the face of volatile supply chains. The company’s future now depends on whether it can monopolise server racks in data centres as effectively as it used to monopolise the desks of corporate employees. If margins on AI hardware fail to compensate for rising component costs, analysts’ current enthusiasm may prove premature.
