While tech giants' capital expenditures skyrocket to fund AI development, Apple's has declined. The company strategically sidesteps the costly race to build foundation models by partnering with Google. It will integrate Gemini into its products, letting Google bear the immense infrastructure and training costs.

Apple's inability to ship its own cutting-edge AI model has paradoxically become a strategic advantage. Instead of bearing the immense cost of foundation model development, they can now integrate best-in-class third-party models onto their dominant hardware ecosystem, a position Mark Gurman calls 'falling ass backwards into it.'

While other tech giants are massively increasing capital expenditures to build AI data centers, Apple's CapEx is down. This reveals a deliberate strategy to avoid the high costs of training foundation models by integrating third-party AI, like Google's Gemini, into its products.

Apple is letting rivals like Google spend billions on building AI infrastructure. Apple's plan is to then license the winning large language models for cheap and integrate them into its massive ecosystem of 2.5 billion devices, leveraging its distribution power without the immense capital expenditure.

Apple's dominant hardware and App Store ecosystem allow it to generate over $1B in annual revenue from AI app fees. This strategy outsources the massive capex and R&D risk to AI labs like OpenAI, creating a high-margin business while they refine their own on-device AI plan.

Apple is avoiding massive capital expenditure on building its own LLMs. By partnering with a leader like Google for the underlying tech (e.g., Gemini for Siri), Apple can focus on its core strength: productizing and integrating technology into a superior user experience, which may be the more profitable long-term play.

Instead of bearing the massive capital expenditure of building foundational AI models, Apple is paying Google billions to use its Gemini technology as the underpinning for Siri. This mirrors its long-standing hardware strategy with Foxconn: let a partner handle the capital-intensive part of the process.

Apple is successfully navigating the AI race by avoiding the massive expense of building foundational models. Instead, it's partnering with companies like Google for AI capabilities while focusing on its core strength: selling high-margin hardware. This allows Apple to capture the end-user without the costly infrastructure build-out of its rivals.

Apple is consciously avoiding the massive CapEx race dominated by Google and Meta. By partnering for heavy-duty AI processing (like with Google Gemini and NVIDIA GPUs), Apple focuses on on-device and privacy-centric AI, making it a potentially capital-efficient player that avoids owning the most expensive parts of the AI stack.