Major retailers like Target are now carefully reconsidering their broad AI adoption strategies.
The reason is a fundamental change in how AI providers bill for their services. We're moving away from predictable, flat monthly subscriptions—like a Netflix plan—to a metered, usage-based model. Think of it like your electricity bill; you pay for exactly what you use. Companies like OpenAI and Anthropic are now charging based on 'tokens,' which are small pieces of text. Every question you ask an AI and every answer it gives consumes tokens, and each token has a price.
This shift turns a predictable software cost into a variable, and potentially very large, operational expense. For a company like Target, which operates on a slim 4.5% operating margin, uncontrolled costs are a serious concern. A scenario with 10,000 employees using advanced AI heavily could rack up bills of nearly $1.5 million per month. This isn't just an IT issue anymore; it's a topic for the CFO and the highest levels of leadership.
So, what caused this sudden change? First, the AI providers themselves changed the rules. In April 2026, OpenAI launched its powerful GPT-5.5 model with higher per-token prices, and Anthropic began moving its enterprise clients off bundled plans to this new pay-as-you-go model. This immediately changed the cost equation for any company using these advanced tools.
Second, cost transparency has improved dramatically. Tools like AWS Bedrock's cost attribution now allow executives to see exactly how much each department, application, and even AI model is costing them. When you can track spending with such detail, you're naturally driven to control it more tightly. This visibility is forcing conversations about governance and ROI.
Finally, the underlying hardware costs for running AI are soaring. Building and maintaining the data centers filled with powerful chips from companies like Nvidia is incredibly expensive. AI providers are passing these high and rising costs onto their customers through usage-based pricing, as it protects them from the risk of users consuming vast, unpredictable amounts of resources for a single flat fee.
As a result, the initial excitement of deploying AI everywhere is being replaced by a more mature, strategic approach. Companies are now focused on 'intentional' AI integration, implementing cost guardrails, and ensuring every AI-powered task delivers clear value. The future of enterprise AI isn't just about capability; it's about cost-effective, sustainable implementation.
- Glossary
- Token-based billing: A pricing model where customers are charged based on the number of 'tokens' (small units of text, roughly 4 characters) their AI model processes.
- Agentic workflows: AI systems that can perform complex, multi-step tasks autonomously, often consuming a large number of tokens as they reason and execute actions.
- Operating Margin: A measure of profitability that indicates how much profit a company makes from its core business operations, expressed as a percentage of revenue.