The world's richest tech giant barely invests in AI. Here is why.

Apple is the most profitable technology company in the world. It generates nearly $144 billion in quarterly revenue at a 29 percent profit margin. It produces $54 billion in operating cash flow per quarter and returns all of it to shareholders. By almost every metric, Apple is winning.

Yet, in artificial intelligence the defining technology competition of our time Apple appears to be losing.

While Microsoft, Google, Amazon, and Meta are collectively spending some $650 billion on AI infrastructure this year, Apple is spending only $14 billion. While competitors race to build larger models and more powerful systems, Apple licenses Google's Gemini to power its revamped Siri. While the industry is obsessed with training the next frontier model, Apple focuses on AI that runs on your device rather than in the cloud.

This looks like a strategy born of weakness. It might well be a strategy born of clarity.

The spending paradox

The numbers tell a striking story. In 2024, Apple spent approximately $14 billion on AI and related infrastructure. For comparison: Microsoft invested $94 billion, Meta $70 billion, Google's AI capex exceeded $60 billion, and Amazon's infrastructure spending approached $60 billion as well.

Yet, Apple remains more profitable than all these companies. Its operating margin stands at 29 percent. Microsoft's margin is around 35 percent, but Microsoft spends nearly seven times as much capital to achieve that. Google's margin sits around 20 percent. Meta's margin is approximately 32 percent, but the company is burning cash on infrastructure without a clear path to AI-driven returns.

This is not a company lagging due to financial constraints. This is a company making a deliberate choice.

The device-first bet

Apple's strategy rests on a fundamental thesis: the future of consumer AI lies on the device itself, not in the cloud.

This means processing takes place on your iPhone, iPad, or Mac instead of on distant servers. Privacy as a default your data does not travel to a data center. Lower latency, offline capability, and less dependence on connectivity. And lower operational costs.

For years, this was dismissed as a limitation. Apple couldn't build large language models, so it had to sell on-device processing as a virtue. But the economics are shifting. As prices for cloud-based AI collapse Anthropic cut prices by 67 percent, Google slashed rates by 70 to 80 percent, OpenAI has repeatedly lowered costs with successive models the commodity nature of frontier models becomes harder to ignore. If everyone can license powerful models cheaply, the advantage shifts to distribution, integration, and user experience.

Apple has distribution. It has 2 billion active devices worldwide. It has an ecosystem that works. It has users who tolerate friction because everything else integrates seamlessly.

It does not have the best AI.

The Siri problem and why it may not matter

Siri is objectively weak by modern standards. It cannot handle complex reasoning. It struggles with multi-step tasks. It does not understand context the way ChatGPT does. Compared to Google Assistant or Alexa, Siri feels like technology from five years ago.

Apple knows this. In January 2026, the company announced that a rebuilt Siri would appear in the spring of 2026, powered by Google's Gemini. Instead of investing the resources to build world-class AI capabilities internally, Apple licenses them from a competitor.

This is the clearest possible declaration: we choose not to compete on raw AI capability.

Yet, consumers are not abandoning iPhones. Apple continues to sell record numbers of devices. Its market capitalization has crossed the $4 trillion mark. Users tolerate a mediocre Siri because they value the ecosystem as a whole more than an exceptional assistant.

Even more intriguingly: Apple earned nearly $900 million from generative AI apps in the App Store in 2025. Three-quarters of that revenue came from ChatGPT. Apple profits from competitors' AI while deliberately underinvesting in its own.

The licensing game

Apple's partnership with Google is telling. Instead of building its own large language models, Apple licenses Gemini. This gives Apple users access to genuine AI capability without the multi-billion dollar infrastructure bill. It gives Google an entry point into Apple's ecosystem. It costs Apple a licensing fee but protects the balance sheet.

This is a bet that in a commodity market, integration and distribution count for more than raw capability. That a good-enough assistant deeply integrated into your device is worth more to users than a brilliant assistant living in the cloud. That Apple's strength making devices that work together seamlessly is a better defensive line than building the largest models.

Why other tech giants cannot make this choice

Microsoft, Google, Amazon, and Meta are trapped in a different logic. They are cloud companies. Their revenue models depend on data center utilization, on users spending time in their ecosystems, on processing power sold as a service. They must build large models because their business model demands it. They must invest in infrastructure because that infrastructure is their product.

Apple's business revolves around devices and services. It does not need to own the AI. It needs to integrate the AI well enough so that users prefer Apple devices.

This is why Apple spends a fifth of what Microsoft spends and yet remains significantly more profitable. This is why Apple's 'restraint' emerges as an advantage now that investors are beginning to doubt whether the trillion-dollar AI infrastructure build-out will ever deliver a return.

The open question

None of this guarantees that Apple's strategy will work. The rebuilt Siri may still feel inferior. Users may ultimately demand better AI and switch platforms. On-device processing may prove inadequate for the tasks that truly matter.

But for business leaders following this competition, Apple's approach offers a different framework. In an era where everyone assumes the solution to competitive pressure is to spend more, Apple asks whether spending differently might be smarter. In an era of AI arms races, Apple chooses selective partnerships over vertical integration.

Whether that turns out to be visionary clarity or shortsightedness will become clear in the next two years. What is clear now: Apple is playing an entirely different game and winning on the scoreboards that truly matter.

Want to get started with AI yourself? Check out our e-learning AI Agents or AI workshops for teams.

// About the author

Remy Gieling

Mede-oprichter, AI-expert & bestseller-auteur

Tech-expert (1988) gespecialiseerd in kunstmatige intelligentie en mede-oprichter van ai.nl, The Automation Group, Proxies en eBrain.ai. Oud-hoofdredacteur van diverse zakenmerken en daardoor een geoefend verteller op het podium en in de media. Verzorgt jaarlijks 150+ AI-keynotes in binnen- en buitenland en is gastdocent aan Nyenrode. Co-auteur van zeven boeken, waaronder 'Handboek AI Strategie' en 'AI Agents', en bekend als presentator op radio en RTL Z. Reist langs de labs van OpenAI, Nvidia en Tencent en vertaalt de nieuwste doorbraken naar inzichten die leiders direct kunnen toepassen.

LinkedIn