The news: The latest multibillion-dollar AI deals are a reflection of the massive computing power needed to build AI products and services—and 2025 has shown just how steep the costs can get. We expect price increases to trickle down to brands, resulting in more expensive access to AI tools and services.

Nvidia CEO Jensen Huang recently projected that $3 trillion to $4 trillion will be spent on AI infrastructure by the end of the decade, per TechCrunch. The pressure to monetize AI is spurring a deluge of billion-dollar deals, some of which create funding loops between industry leaders that could take years to pay off while driving up costs for customers in the interim.

Clouds, chips, and cash fuel the AI race: It’s clear that the biggest tech, infrastructure, and AI players are looking to partnerships and deals to solidify their dominance. The trend could leave smaller companies and AI startups struggling to catch up.

• Microsoft. Its 2019 $1 billion Azure-backed bet on OpenAI, which grew to $14 billion before loosening, set the template.
• Amazon. The $8 billion Anthropic deal—and Google Cloud’s smaller tie-ups—revealed the formula: Bind an AI model to your stack and scale in tandem.
• Oracle. Its $30 billion cloud services deal in June with OpenAI, later dwarfed by a $300 billion, five-year commitment announced in September, makes the database giant one of the biggest players of the AI arms race.
• Meta. The social media firm has taken a different route, sinking $600 billion into US infrastructure by 2028 to own both sides of the compute-AI model equation, allowing it to set its own pricing for AI services.

Our take: As hardware giants, cloud providers, and model labs form alliances, the landscape is tilting toward closed ecosystems with higher barriers to entry and less choice for brands.

Compute-driven services will get more expensive to recoup investments, and fewer but more integrated platforms will dictate the tools available. Brands should stay agile—diversify vendor bets where budgets allow, test across ecosystems, and align with the stacks most relevant to audiences and workflows.

Failing to anticipate and respond to this hardware–cloud–AI consolidation could leave brands locked into costly, limited ecosystems with little room to maneuver.