As the Price of Intelligence Collapses, Agentic AI Is Reshaping the Workforce

Last Wednesday, WiseTech Global became the first Australian company to announce mass AI-related job losses. The logistics software business told 2,000 staff they were being let go as part of an AI transformation. Two days later, Jack Dorsey’s Block Inc announced it would cut roughly 4,000 of its 10,000 employees.

“We’re not making this decision because we’re in trouble. Our business is strong,” Dorsey told his remaining staff. “But something has changed. We’re already seeing that the intelligence tools we’re creating and using, paired with smaller and flatter teams, are enabling a new way of working which fundamentally changes what it means to build and run a company.”

“If you need to ring us, then we’ve done something wrong.” — Paul Tyler, CEO of Superloop

The Rise of Agentic AI

We are witnessing a fundamental shift in how AI is deployed. Instead of simply asking chatbots questions, companies are now “hiring” AI agents—also known as agentic AI—to perform actual work.

These agents are pieces of software designed to apply AI reasoning and data to specific business tasks: handling customer queries, developing marketing plans, processing legal documents, and more. They sit atop large language models like ChatGPT, Gemini, and Claude, purchasing “intelligence” from them as needed.

The analogy is striking: the LLM is the brain, and the agent is the nerves, bones, and muscle putting the brain’s thinking to work.

The Collapsing Price of Intelligence

The economics driving this transformation are staggering. According to industry data, the price of middling intelligence—measured on an Artificial Analysis index—has fallen from $30 to just 3 cents over the past two years. That’s a 99.9% price collapse.

Businesses now switch between LLM providers freely, buying the cheapest tokens available. Tokens—the currency of machine intelligence—have become a commodity. One thousand tokens equals roughly 750 words of processing power.

When an agent tackles a complex problem, it pings the LLM repeatedly. Each interaction consumes “Input Tokens” (the instruction) followed by “Output Tokens” (the decision or written product). With prices this low, companies can deploy AI agents at scale for fractions of what human labor costs.

“We may no longer be the most intelligent things on Earth. For the rest of our lives, we’ll all be sleeping next to a tiger.” — Noah Smith, Economics Blogger

China’s Disruptive Entry

As it did in manufacturing, China is now undercutting US AI models. DeepSeek—while not quite matching the quality of American models—is “usually good enough” and putting immense pressure on industry pricing by offering PhD-level reasoning for free.

This threatens to wipe out the standard $20 subscription tiers and is forcing American companies to focus on premium “agentic” versions priced at $200-300 per month. These premium tiers don’t just answer questions—they do the work.

DeepSeek’s pricing for developers and enterprises is pay-as-you-go: 28 cents per million tokens for chat and reasoning modes, or $2 per million for the advanced DeepSeek-V3.2-Exp model.

What This Means for Workers

The implications extend far beyond tech companies. Paul Tyler, CEO of Australian internet provider Superloop, notes that almost all customer interactions are now handled by AI. His company has achieved headcount reductions through natural attrition rather than layoffs—but the result is the same: fewer humans needed.

As OpenAI CEO Sam Altman recently observed when defending AI resource consumption: “It takes like 20 years of life and all of the food you eat during that time before you get smart. And not only that, it took the very widespread evolution of the 100 billion people that have ever lived… to produce you.”

Now that intelligence can be manufactured and sold by the million tokens, the economic calculus for employers has fundamentally changed.

The Road Ahead

An entire retail and wholesale ecosystem for intelligence has emerged from nothing in about a year—yet most people don’t yet understand what’s happening. The LLM providers are quickly becoming low-margin utilities, even as stock markets price them as high-growth tech companies.

What makes this product different from all others is that what’s being sold is, quite literally, intelligence. It’s called artificial to distinguish it from the human variety—it doesn’t yet include empathy and other human emotions—but functionally, it’s the same thing.

The question now isn’t whether AI will transform work. It’s how quickly, and who gets left behind.

This article was reported by the ArtificialDaily editorial team. For more information, visit ABC News.