Apple keeps AI spend flat

Apple is approaching artificial intelligence with less visible spending growth than the model labs that dominate the sector’s capital race. A May 22 AppleInsider report said Apple’s AI spend has remained comparatively flat even as the company continues to build AI into hardware, software and services. The same report contrasted Apple’s posture with OpenAI, which it said is losing $1.25 for every $1 of revenue. That contrast helps explain why Apple’s AI strategy looks different from the cloud-heavy buildouts associated with standalone model companies. Apple has emphasized shipping intelligence inside existing products rather than framing AI as a separate business line, according to the AppleInsider report. Apple’s next major venue for discussing software and developer tools is WWDC26, which runs June 8 through June 12, Apple said. (appleinsider.com) ### Why does “flat spend” matter in Apple’s case? AppleInsider reported on May 22 that Apple continues to benefit from AI “without all of the spend,” describing a strategy that differs from companies funding large standalone model operations. The report did not present Apple as exiting AI investment; it presented Apple as keeping spending more restrained while embedding features into its installed base. (appleinsider.com) Apple’s business gives it other levers. The company controls devices, operating systems, app frameworks and services, which lets it place AI features where it already has distribution. That setup can reduce the need to chase scale in the same way as companies whose products depend primarily on remote model access. This is an inference drawn from Apple’s product structure and the AppleInsider report’s description of an integration-first approach. (appleinsider.com) ### What does the OpenAI comparison show? AppleInsider’s most arresting figure was its claim that OpenAI is losing $1.25 for every $1 of revenue. In the context of the story, that number was used to show how expensive frontier-model economics can be when training and serving large systems at scale. For Apple, the comparison is less about model quality than cost structure. (appleinsider.com) If a company can deliver consumer AI features through a mix of on-device processing and selective cloud use, it may avoid some of the recurring inference expense that burdens pure cloud AI providers. That cost reading is an inference supported by the AppleInsider report and by separate reporting on local model use on Apple hardware. ### How does local inference on M5 Max hardware fit in? Geeky Gadgets reported on May 23 that Apple’s M5 Max MacBook Pro, with 128GB of unified memory and 40 GPU cores, can run large language models locally without relying on external servers. The publication framed local use as a way to eliminate cloud API costs for some workloads. That does not amount to an Apple product announcement, but it does illustrate the hardware case for local AI. (appleinsider.com) If more tasks can run on-device, Apple can route some requests away from the cloud, which can help with latency, privacy and operating cost. That hybrid design — local for lighter or private tasks, cloud for heavier workloads — is an inference from the M5 Max report and AppleInsider’s spending story. (geeky-gadgets.com) ### What should readers watch next? Apple said WWDC26 will run online from June 8 to June 12, with a special event at Apple Park on June 8. The conference will give developers a first look at Apple’s latest software, tools and features. June 8 is the next clear date for testing whether Apple keeps presenting AI as a tightly integrated product feature rather than a spending race. (appleinsider.com) Any new disclosures about Apple Intelligence, Siri, developer frameworks or on-device model capabilities are likely to be central to that update. (apple.com 1) (apple.com 2)