Apple reported its best March quarter in history on Thursday, posting revenue of $111.2 billion, up 17 per cent year over year, with net profit of $29.6 billion. iPhone revenue reached a March quarter record of $58 billion, up 22 per cent, driven by what Tim Cook called “extraordinary” demand for the iPhone 17, which he described as the most popular launch in the company's history. Sales in Greater China rose 28 per cent to $20.5 billion, outpacing every other region. Services revenue hit an all-time high of $31 billion. The company authorised a new $100 billion share buyback. And the stock moved barely at all, because the market has learned that Apple's best quarters tend to arrive alongside a warning. This time, Cook told analysts that “significantly higher memory costs” would begin affecting margins in the June quarter, with the impact accelerating through the rest of the year. The best quarter in Apple's history came with a memo that the next one will be more expensive to produce.

The strategy that should not have worked

Apple's results are a rebuke to the prevailing logic of the AI era. While Meta has raised its 2026 capital expenditure guidance to $145 billion, Amazon to $200 billion, and Alphabet to $190 billion, Apple's total capex remains a fraction of those figures. The company has not built its own large language model. It has not constructed a fleet of AI data centres. It has not hired 10,000 GPU engineers or entered bidding wars for compute capacity. Instead, Apple chose to partner with OpenAI and Google, integrating their AI models into its devices through Apple Intelligence, the company's AI framework that routes tasks to on-device processing where possible and to cloud-based models from partners where necessary. Cook described the approach on Thursday's earnings call as AI that is “respectful of user privacy” and “not a standalone feature, but an essential and intuitive part of our devices.” The positioning is deliberate. Apple is not selling AI as a product. It is selling devices that happen to use AI, made by companies that spent hundreds of billions of dollars developing the models Apple did not have to build.

The financial results suggest the strategy is working. Apple is capturing the consumer benefits of the AI wave, improved photo editing, smarter search, more capable Siri, enhanced app recommendations, without bearing the infrastructure costs that are compressing margins at its peers. The iPhone 17's record demand is partly attributable to the Apple Intelligence features bundled with the device, features that cost Apple a licensing fee rather than a data centre. The MacBook Neo, the $599 laptop that Cook said was seeing “off the charts” demand, set a record for new Mac buyers by putting an A18 Pro chip, the same processor found in the iPhone 16 Pro, into a fanless design aimed at students and budget-conscious consumers. The Neo has already sold out through April, and Apple has doubled its production orders to 10 million units. Intel launched an entire new processor line in response. Apple's competitive advantage in AI is not the AI itself. It is the integration of other companies' AI into hardware that Apple designs, manufactures, and sells at margins its competitors cannot match.

The handover

This was the first earnings call since Apple announced on 20 April that John Ternus, the company's longtime hardware chief, would succeed Cook as chief executive on 1 September, with Cook moving to the role of executive chairman. Ternus spoke briefly to financial analysts for the first time, promising “financial discipline” and alluding to a product roadmap he described as “the most exciting time in my career at Apple to be building products and services.” Cook used the call to praise his successor, saying “I know he will push us to go further than we think is possible.” The tone was gracious and scripted. The substance was a signal: Ternus is a hardware executive inheriting a company whose next chapter depends on software and services.

The challenge Ternus faces is whether Apple's partnership-based AI strategy can survive the next phase of the technology's development. Apple's bet on Ternus comes at a moment when its competitors are building proprietary AI systems that are increasingly integrated into their own hardware and services ecosystems. Google's Gemini models power search, Android, Workspace, and Cloud. Meta's Muse Spark is embedded across Facebook, Instagram, WhatsApp, and the Ray-Ban glasses. Microsoft's Copilot is woven into Windows, Office, and Azure. Each of these companies controls the full stack from model to interface. Apple controls the interface but relies on partners for the model, which means its AI capabilities are ultimately constrained by what OpenAI and Google are willing to provide and at what price. Cook's Apple Intelligence update, promising that AI will be more deeply integrated into Siri later this year, suggests the company knows this dependency needs to evolve. Whether Ternus will build, buy, or continue to borrow remains the defining strategic question of his tenure.

The others

Apple was not the only company reporting on Thursday. Reddit posted first-quarter revenue of $663 million, up 69 per cent year over year, with advertising revenue growing 74 per cent to $625 million. The platform reported 126.8 million daily active users globally, up 17 per cent, and CEO Steve Huffman said that 200 million people in the United States, more than half the country's population, visit Reddit every week. Huffman said the company's data licensing deals with OpenAI and Google, which pay Reddit for access to its conversations to train their AI models, are valuable but will become even more important as the rest of the internet becomes “optimised for AI.” His argument is that Reddit's human-generated conversations are a scarce resource in a world increasingly flooded with AI-generated content. “At the end of the day, there is no artificial intelligence without actual intelligence,” Huffman told analysts. It is a line that reads as clever branding, but it also captures a real dynamic: as AI models improve, the training data that makes them work becomes more valuable, and Reddit sits on one of the largest repositories of authentic human discourse on the internet.

Roblox, by contrast, demonstrated what happens when a platform's growth collides with its obligations. The gaming company reported first-quarter revenue of $1.4 billion, up 39 per cent, and daily active users of 132 million, up 35 per cent. But its rollout of mandatory age verification, introduced in response to lawsuits from attorneys general in eight US states over child safety failures, slowed new user acquisition and reduced engagement among unverified accounts. CEO David Baszucki told investors that “growth was tempered by greater-than-expected headwinds from our age-check roll out,” and the company cut its full-year revenue guidance, sending shares down 20 per cent in after-hours trading. Roblox has yet to post a profitable quarter since going public in 2021. The age verification experience is a preview of what every platform serving young users will face as regulators worldwide impose stricter requirements: doing the right thing on child safety is expensive, it slows growth, and the market punishes you for it before the regulatory alternative, which is to be sued, punishes you worse.