Tech leaders including Microsoft CEO Satya Nadella are now advocating for cheaper, smaller AI models over expensive premium options, as companies face unexpected cost spikes from usage-based pricing models. What began as a rush to adopt powerful artificial intelligence tools is turning into a cost-cutting reality check, with major corporations discovering that their AI bills are far exceeding initial budgets.

Why Are AI Costs Spiraling Out of Control?

The shift from flat subscription pricing to usage-based pricing has caught many companies off guard. When AI firms started charging by the token, the unit used to measure AI usage, businesses suddenly faced unpredictable and often higher bills. A startup called BlueRock, which helps companies run AI systems safely, reported that customers experienced "a 20% to 30% spike in terms of over-budgeting" immediately after this licensing change.

The problem intensified because AI tasks now require more steps, more data, and longer inputs than companies initially anticipated. Uber provides a stark example: the company burned through its entire 2026 AI budget in just four months after employees rushed to adopt AI coding tools, forcing management to cap usage. Gartner, a major research firm, estimates that AI coding costs will surpass the average developer's salary by 2028, while three-quarters of executives surveyed expect tech budgets to rise this year, with nearly half projecting double-digit jumps.

What Are Tech Leaders Saying About Smaller Models?

Top executives from across the industry are now making the case that businesses don't need the most expensive, powerful AI models for most tasks. Beyond Nadella at Microsoft, leaders like Nikesh Arora from Palo Alto Networks and Brian Armstrong from Coinbase Global have publicly stated that smaller, cheaper models can handle a significant share of corporate needs.

"If you want to win enterprise, you should be forward pricing tokens," said Nikesh Arora, Palo Alto Networks executive, urging AI labs to charge customers today at the lower rates that tokens are expected to command in a few years.

Nikesh Arora, Palo Alto Networks

This represents a dramatic shift from what industry insiders call "tokenmaxxing," a practice where companies encouraged heavy AI usage, treating rising consumption as a proxy for productivity. Now that the bills are arriving, that strategy looks shortsighted.

How Are Companies Adapting to Lower-Cost AI Options?

• Routing Tools: Businesses are turning to AI marketplaces like OpenRouter to assign tasks to the most cost-effective system while reserving premium models for complex work such as coding and advanced reasoning.
• Open-Source Models: Chinese open-source models, particularly DeepSeek, are gaining traction among startups and increasingly among larger enterprises, charging as little as 18 cents per million tokens compared to an average of $4 for top-tier models.
• Multi-Provider Strategy: Companies are following the cloud computing playbook, spreading across multiple providers to find the best fit and price rather than relying on a single expensive vendor.

The shift toward cheaper alternatives is already visible in usage data. Open-source tokens processed on OpenRouter jumped to 65% in June from 34% in January, according to analysis from Citi. The four most popular models on OpenRouter are all Chinese, with DeepSeek holding the top spot.

Chinese models are rapidly closing the capability gap with top U.S. models. BlueRock CEO Harold Byun noted that open-source models "used to be more than a year behind (leading AI models). Now, probably the estimates are they're roughly four months behind. That the gap will continue to close".

Harold Byun

What Does This Mean for OpenAI and Other AI Companies?

The pressure to compete on price could hurt revenue growth for premium AI providers, especially as they prepare for potential initial public offerings (IPOs). OpenAI has been reported to be weighing significant price cuts, including on token usage, in anticipation of similar moves from rival Anthropic.

"There will be a price-war dynamic when it comes to OpenAI and Anthropic as they both duke it out for a 'first to public market' IPO dates," said Christopher Brown, financial adviser in private wealth management at Synovus Securities.

Christopher Brown, Financial Adviser, Synovus Securities

The cost pressure is also reshaping how enterprises view AI capabilities. According to Val Bercovici, chief AI officer at WEKA, a company that helps businesses run AI models faster and cheaper, open-source models are demonstrating that they are "90% as good at 10% of the price." This suggests that businesses don't need to spend premium token costs on every level of effort.

What Are the Remaining Barriers to Cheaper AI Adoption?

Despite the cost advantages, security concerns remain a significant obstacle to enterprise adoption of Chinese AI models, particularly in sensitive industries such as cybersecurity. Analysts expect that instead of a wholesale shift to cheaper alternatives, businesses will follow a diversified approach, using multiple providers based on specific needs and risk profiles.

The broader trend reflects a maturing AI market where cost efficiency is becoming as important as raw capability. As Nadella and other leaders have emphasized, the future of enterprise AI may not belong to the most powerful models, but to the ones that deliver the best value for specific business problems.