Apple has taken a different approach to the AI buildout than hyperscalers like Google and Amazon, avoiding massive capital expenditures on proprietary data centers. According to KIWOOM Securities analyst Seung-hyuk Kim, Apple's nine-month capital expenditure fell 28 percent to $6.8 billion, made possible by a hybrid infrastructure model that combines its own servers with external cloud partners such as Google.
Despite this leaner spending, Apple has acknowledged in a 10-Q filing with the SEC that it faces real risks from supply constraints. The company said turmoil in the NAND and DRAM markets could adversely affect its ability to obtain sufficient components on reasonable terms. It also disclosed that demand for AI and machine learning computing infrastructure has grown substantially, leading to constrained supply, extended lead times, and rising costs. Apple warned that if it cannot secure enough computing capacity, either through its own resources or third-party providers, the functionality, availability, or timing of its products and services could be affected.
These broader supply pressures are already visible in specific products, per Kim. The MacBook Neo has reportedly seen its 2026 shipment tally drop by 40 percent due to a shortage of A18 Pro chips, and the same constraint is expected to affect the upcoming iPhone 18 Pro and iPhone 18 Pro Max. Kim suggests that with the DRAM shortage expected to persist until 2028, rising memory costs could weigh on hardware demand going forward.
To offset this, Kim believes Apple needs to lean more heavily on services and AI features to drive recurring revenue and hardware upgrade cycles. Apple's Services division brought in $30.74 billion in Q3 2026, down slightly from $31 billion in Q2 2026, ending a 14-quarter streak of growth. Kim recommends Apple prioritize Siri AI integrations, potentially bundled with iCloud+ under new pricing tiers, to reinforce this business segment. Apple has also introduced a leasing program allowing customers to save $547 off the $1,099 price of an iPhone 17 Pro over a 12-month term, after which the device can be returned to start a new lease on an iPhone 18 Pro.
On the infrastructure side, Apple's AI and machine learning workloads run through its Private Cloud Compute (PCC) system. Unlike Google and Amazon, which have developed custom in-house AI chips, Apple has shared few details about its own hardware approach. In June, the company announced it had worked with Google and NVIDIA to extend PCC to NVIDIA's chips via Google Cloud, reflecting its continued reliance on external partners even as it seeks to manage rising costs and constrained supply across both memory components and AI computing capacity.