After 15 years, planning to re-enter the server market: How ambitious is Apple Inc. (AAPL.US) about AI computing power? Major banks debate "distant water" versus "immediate thirst"

According to people familiar with the matter cited by tech media outlet The Information, Apple Inc. (AAPL.US) is considering entering the enterprise server market, with plans to adopt its self-developed M-series Ultra chips paired with networking equipment from NVIDIA Corporation (NVDA.US). The report states that Apple Inc. intends to sell servers to AI developers, enterprises, and government clients, with the core focus on enabling customers to run local AI inference on privatized infrastructure. According to the disclosure, the proposed servers will come in two versions: a lower-spec version integrating two M8 Ultra chips still under development, and a higher-spec version equipped with four M8 Ultra chips forming a computing cluster. To address high-speed interconnect between multiple chips, Apple Inc. has held talks with NVIDIA Corporation about adopting its NVLink Fusion interconnect solution (covering switches, chiplets, and accompanying software). Should this materialize, it would mark Apple Inc.'s re-entry into the server market. Between 2002 and 2011, Apple Inc. launched the Xserve server powered by Intel Corporation and IBM chips, but due to insufficient emphasis on the enterprise market and weak customer support, the product never gained traction and was ultimately discontinued in early 2011. Mac sales surge 29% ignites server ambitions However, the market environment today is vastly different from back then. In Apple Inc.'s fiscal 2026 third-quarter earnings report, Mac business revenue surged 29% year-over-year to $10.3 billion, leading all product lines including iPhone and services in growth rate. It is understood that major AI labs such as OpenAI have purchased tens of thousands of Mac minis and Mac Studios for training AI agents through reinforcement learning, and Anthropic has also rented Mac minis from Amazon.com, Inc.'s cloud computing armsuch robust demand even led to severe shortages of these models at one point. Yet the report also notes that even if Apple Inc. advances the project, the new products would not be available until 2029, and the project could still be cancelled. Although details remain undecided, Apple Inc.'s new CEO John Ternus expressed support as early as over a year ago when the project was initiated (when he still headed the hardware engineering department)the plan opens the door for Apple Inc.'s self-developed chips to enter the AI computing power market. Following the news, Apple Inc.'s stock price rose slightly by 0.7% in late morning trading on Wednesday, while NVIDIA Corporation's stock price rose more than 2% at one point during the same day's trading. It is worth noting that CICC pointed out in its earlier commentary on Apple Inc.'s 3QFY26 earnings that quarterly revenue of $109.417 billion (+16% YoY) and net profit attributable to shareholders of $29.789 billion (+27% YoY) both exceeded expectations, with Mac revenue of $10.352 billion (+29%) being the biggest contributor; it maintained its FY2026/27 earnings forecasts unchanged and raised its target price by 10% to $340, citing "accelerating AI progress and rising valuation center." If the server plan materializes, it would undoubtedly further reinforce this "self-developed chip value re-rating" logic. Regarding the server news itself, multiple institutions believe the relationship between "distant water" and "immediate thirst" should be viewed calmly: the 2029 launch window means the project will contribute virtually nothing to Apple Inc.'s FY2026-27 earnings forecasts, does not constitute a short-term performance catalyst, and is better viewed as a "strategic call option." Some institutions calculate that the truly critical underwriting question iswhether Apple Inc. can sell differentiated privatized AI computing power at a premium (leveraging unified memory architecture, energy efficiency ratios, and privacy security advantages), rather than becoming just another buyer purchasing expensive computing power. In other words, before obtaining confirmed customer commitments, capital expenditure, or pricing model disclosures, the news is insufficient to support a significant valuation re-rating. Trillion-dollar track competitive landscape: Giants all around, what can Apple Inc. stand on? Goldman Sachs Group, Inc. estimates that driven by the AI wave, the global AI server market size will reach $1.24 trillion by 2030. This track is already extremely crowded, and players' performance momentum remains red-hot: Dell Technologies, Inc. Class C: Latest quarterly revenue of $46.97 billion (+57.8% YoY), single-quarter AI server orders reaching $60.9 billion, backlog at period-end reaching $95 billion, stock price up more than threefold year-to-date; HPE: Latest quarterly revenue of $12.21 billion (+32.7%), surged as much as 137% year-to-date driven by AI and networking businesses, only experiencing a deep correction in mid-September due to a rating downgrade; Super Micro Computer, Inc. (SMCI): Latest quarterly revenue of $11.12 billion (+93.2%), order backlog also hitting record highs. Apple Inc.'s differentiated approach lies in not directly competing with NVIDIA Corporation's GPU clusters for the large model training market, but instead the AI inference segment, particularly "privatized deployment" niche scenariostargeting enterprise and government clients who want to run models on their own servers and are highly sensitive to data privacy. This positioning naturally aligns with Apple Inc.'s hardware energy efficiency advantages and unified memory architecture, and is also an area where NVIDIA Corporation is currently relatively weak but rapidly catching up. However, Apple Inc.'s shortcomings are equally clear. Servers are essentially a low-margin, capital-intensive, service-heavy B2B business, vastly different from Apple Inc.'s customary high-margin consumer electronics playbook; enterprise buyers value software ecosystems, remote management, compliance certifications, and full lifecycle supportwhich is precisely the core lesson from the Xserve era's failure. Additionally, Apple Inc.'s self-developed interconnect technology currently used for internal Private Cloud Compute has speed and cost bottlenecks under hyperscale deployment, which is also the direct reason for seeking NVIDIA Corporation's NVLink Fusion. Industry chain and landscape projection: NVIDIA Corporation may be the "hidden winner" If the cooperation ultimately materializes, the market widely believes NVIDIA Corporation is the more certain beneficiary. Networking equipment currently accounts for approximately 10%-15% of total AI data center hardware costs, and the networking business already contributes about one-fifth of NVIDIA Corporation's data center revenue. Since NVLink Fusion's launch in May 2025, it has attracted partners including MediaTek, Marvell, Fujitsu, and Qualcomm, and Amazon.com, Inc.'s cloud computing arm announced last month that its next-generation AI servers will expand adoption. If Apple Inc. joins, it would be the platform's most significant customer to datethis not only means a new revenue source but also validates NVIDIA Corporation's ecosystem strategy of "even if customers use self-developed chips, they still need my interconnect," thereby further fortifying its moat against the open standards camp. Intriguingly, Apple Inc.'s interconnect route choice: Apple Inc. is simultaneously a board member of the UALink alliance (an open interconnect standards organization formed by AMD, Intel Corporation, Alphabet Inc. Class C, Microsoft Corporation, and others, with over 65 members), and if it ultimately leans toward NVIDIA Corporation's proprietary solution, it would effectively be choosing the latter between "openness" and "performance," and the industry interconnect standards battle may face a key turning point. Also noteworthy: some Apple Inc. engineers involved in the project believe NVIDIA Corporation has the best interconnect technology currently available, but the project could also proceed without adopting NVIDIA Corporation technology. On the risk side, industry-wide memory chip shortages continue to push up hardware costs, and Apple Inc. would need to lock in component supply years in advance if it wants to deliver servers at scale; meanwhile, the gap between its AI software ecosystem (such as the MLX framework) and CUDA, as well as core talent attrition during the year, are unavoidable challenges. Conclusion For Ternus, who has just completed the CEO transition, the server project is both a natural extension of his hardware engineering background and potentially a strategic gamble that could define his tenure's legacy. In the short term, the news provides new imaginative space for Apple Inc.'s AI narrative, and combined with Mac business growth exceeding expectations, sentiment is relatively warm; but over a longer cycle, whether the 2029 product can carve out a gap in the fiercely competitive inference market depends on whether Apple Inc. can shore up enterprise-grade capabilitiesits biggest shortcoming. For investors, rather than chasing rumor-driven rallies, it is better to closely watch three signals: whether data center capital expenditure is booked, whether anchor customers materialize, and the final choice of interconnect technology route.