Reports say Apple is seeking to acquire a chip company to accelerate its AI server capabilities.

Apple is hitting the accelerator on its AI server chip acquisition strategy. On July 15, according to The Information, Apple is actively seeking chip company acquisition targets and has contacted investment banks and several semiconductor startups over the past few months to evaluate potential deals. Previously, Apple’s self-developed next-generation AI server chip Baltra was delayed, and the current M2 Ultra has reached its performance ceiling for AI tasks in data centers, forcing Apple to outsource related heavy workloads to Nvidia infrastructure hosted on Google Cloud. In January this year, Apple spent nearly $2 billion to acquire the Israeli AI startup Q.ai, marking the company’s second-largest acquisition to date, second only to the $3 billion acquisition of Beats Electronics in 2014. M2 Ultra Performance Peaks, Baltra Delay Fuels Computing Power Anxiety Apple’s current AI data center computing power relies on the M2 Ultra chip, but as AI workloads scale up, its internal servers have hit a performance bottleneck. According to The Information, Apple’s next-generation AI server chip Baltra was originally scheduled to debut this year but has been delayed. Before Baltra is delivered, Apple has to outsource part of its heavy AI workloads to external infrastructure. These tasks currently run on Nvidia chips hosted by Google Cloud. This situation means that, as AI models continue to iterate rapidly week by week, Apple’s core AI infrastructure faces a dual dependence on external suppliers. Of note, previous Morgan Stanley research disclosed that Apple is making large-scale commitments to TSMC's SoIC advanced packaging capacity for Baltra—order volumes reaching 36,000 wafers in 2026, and surging to 60,000 wafers in 2027. Transaction Size and Potential Targets Still Unclear Apple’s January acquisition of Q.ai marked a turning point in its acquisition strategy. Q.ai focuses on technology for decoding speech via facial micro-movements. Its nearly $2 billion valuation indicates that Apple is no longer constrained by its former conservative approach to strategic acquisitions. The logic behind this deal lies not only in acquiring the technology itself but in proving a path: when internal R&D progress falls short of expectations, external acquisitions can quickly fill capability gaps. Now, this logic is being applied to the AI server chip field. According to reports citing insiders, Apple has recently held talks with startup PrismML, which specializes in model compression technology. PrismML claims it can compress 27 billion parameter large models to run locally on iPhone without loss of performance, precisely the key capability needed for Apple’s device-side AI vision. Currently, Apple has not publicly disclosed any specific acquisition targets, nor reveals the scale of potential deals. Related contacts remain in the early evaluation stage. Worth noting, the competitive landscape in AI server chips is highly concentrated, with Nvidia occupying a dominant position. If Apple wants to substantially boost its own AI computing power through acquisitions, the required targets face high thresholds for technological maturity and strategic alignment. Risk Disclaimer The market has risks, investment requires caution. This article does not constitute personal investment advice and does not take into account individual users’ special investment objectives, financial situations, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Investments made accordingly are at their own risk.