Apple has picked Google's Gemini model to power the next generation of Siri, a move that deepens the iPhone maker's reliance on a single AI provider. The deal comes as Alphabet, Google's parent company, announces a $185 billion capital expenditure plan for AI infrastructure — a figure that underscores the sheer scale of investment needed to stay competitive in the field. But the partnership also highlights growing concerns about centralization in the AI ecosystem, fueling interest in decentralized alternatives.
Why Apple chose Gemini
Apple's decision to adopt Gemini for Siri marks a significant shift. The company had previously relied on its own, smaller language models for on-device processing, but the complexity of tasks like summarization and multi-step reasoning required a more powerful backend. Rather than build its own frontier model from scratch — a process that can cost billions and take years — Apple opted to license Google's technology. The arrangement lets Apple offer advanced AI features without the massive infrastructure investment, while Google gains a high-profile customer for its cloud AI services.
Alphabet's $185 billion bet
Alphabet's planned $185 billion in capital expenditures is largely directed at building and operating data centers capable of training and running large language models. That spending dwarfs what most other companies are committing to AI infrastructure. The sheer scale of the investment creates a high barrier to entry, meaning only a handful of tech giants can realistically compete at the frontier of AI development. This concentration of resources is a key driver of the centralization that critics warn about.
The centralization debate
When a small number of companies control the most advanced AI models, the risks multiply. A single point of failure — whether technical, regulatory, or strategic — could affect billions of users. Apple's reliance on Google's Gemini means that any disruption at Google could ripple across Apple's ecosystem. Similarly, Alphabet's massive bet on AI infrastructure gives it outsized influence over the direction of the technology. Critics argue that this concentration of power undermines the open, decentralized ethos that originally drove internet innovation.
Decentralized AI gains traction
The Apple-Google partnership has intensified interest in decentralized AI solutions. Projects that distribute model training and inference across many independent nodes — rather than centralizing them in a few corporate data centers — are attracting more attention and investment. Proponents say decentralized AI can reduce single points of failure, lower costs for smaller developers, and give users more control over their data. Whether these alternatives can match the performance of centralized giants remains an open question, but the debate is no longer theoretical.
Regulators in the US and Europe are already examining the competitive implications of AI partnerships like this one. The European Commission has signaled it may scrutinize deals between Big Tech firms that could entrench market power. Meanwhile, a growing number of open-source AI projects are vying for developer mindshare, hoping to prove that decentralized models can be both capable and trustworthy. The next few months will show whether the market — or regulators — push back against the centralization trend.
