When Thomas Edison built the first electrical systems, they worked—but they didn’t scale because direct current (DC) couldn’t travel more than a mile. Nikola Tesla solved that with alternating current (AC). Same electricity—different system—and suddenly power could go the distance. AI cost is hitting the
This week, stories in The New York Times and The Wall Street Journal highlighted something that’s been quietly building inside companies: employees are deploying AI agents that generate massive volumes of tokens—and massive, unexpected costs. The assumption behind most AI systems is simple: tokens approximate usage, and usage
By Alan Jacobson, AI Economics Strategist This week, stories in The New York Times and The Wall Street Journal highlighted something that’s been quietly building inside companies: employees are deploying AI agents that generate massive volumes of tokens—and massive, unexpected costs. The assumption behind most AI systems is
Agentic AI has destroyed the SaaS seat-based subscription model By Alan Jacobson, AI Economics Strategist This week, stories in The New York Times and Wall Street Journal highlighted something that’s been quietly building inside companies: employees are deploying AI agents that generate massive volumes of tokens—and massive, unexpected
Across 9 public cloud companies, AI-driven cost of revenue is growing faster than revenue itself — with divergence of +300 to +3,700 bps and early margin compression. In several cases, cost growth is 2–5x revenue growth. The AI Cost-Revenue Divergence (ACRD) detection framework compares revenue growth, cost-of-revenue growth, and
By Alan Jacobson, AI Economics Strategist AI must work as a business. Today’s models don’t. Better models don’t fix the economics. Models only improve output — not cost structure. And neither does scale — because unlike search, social, ecommerce, and cloud, AI costs scale with usage instead of compressing
By Alan Jacobson, AI Economics Strategist If AI costs are rising faster than revenue, the problem isn’t the model — it’s how compute is provisioned, governed, and billed. Al is the first spend category where execution is implicitly authorized and billed after the fact. What follows makes Al behave
By Alan Jacobson, Systems Architect & Analyst Can LLMs “scale” their way to profitability by charging a flat rate of $20/month for unlimited compute? * If every man, woman and child in America pays $20/month = $81.6B/year * Capex depreciation: $50B/year * Staff: $11.25B/year * Cost of electricity
By Alan Jacobson, Systems Architect & Analyst The first wave of AI adoption is over. Not because the technology failed, but because users hit its limits faster than the industry expected. Models forgot. Systems hallucinated. Outputs couldn’t be trusted. And when reliability drops, adoption at scale is imp. That
By Alan Jacobson, AI Economics Strategist A recent passage from The New York Times describes a structural failure in modern risk detection: “Our models for detecting risk look at prices, volatility and correlations… They have no instruments for reading a grid failure, a drought or a severed supply chain. By
By Alan Jacobson, AI Economics Strategist 1. AI companies talk endlessly about revenue growth — is AI cost rising faster than AI revenue? 2. If an AI system rehydrates context on every turn and unit cost rises with usage, how is that scaling rather than a cost curve pointing the wrong
By Alan Jacobson, AI Economics Strategist Most businesses run on a simple premise: * You approve your costs before you spend the money. * You budget for them, so you can pay the bill when it arrives. * And you control your costs before they become more than you can afford. It'
By Alan Jacobson, AI Economics Strategist Everyone is debating whether LLMs are accurate enough. That’s the wrong question. The deeper issue isn’t just correctness. It’s predictability — of both output and cost. Two different kinds of systems At a high level, there are two ways software behaves. 1)
By Alan Jacobson, AI Economics Strategist Everyone is focused on what AI can do. Almost no one is focused on what AI costs to run. That’s the problem. The hidden break in the SaaS model SaaS worked because the economics were simple: * build once * sell many times * cost doesn’
By Alan Jacobson, AI Economics Strategist Microsoft’s decision to double down on seat-based pricing for AI tools may feel familiar to enterprise buyers. But familiarity does not make it correct. The company’s new E7 bundle—combining Office 365, Copilot, security software and new agent-management tools—extends the traditional
By Alan Jacobson, AI Economics Strategist * The hare is fast, but every step burns energy. * The tortoise is steady, and keeps moving forward. The problem isn’t speed. It’s cost per step. The core assumption behind SaaS is simple: build once, sell many times, and margins expand as revenue
By Alan Jacobson, AI Economics Strategist AI is compressing margins at 25 companies — and now even OpenAI is pulling back: * OpenAI is retreating from “everything everywhere.” * Microsoft’s CEO is stepping into Copilot. * Meta is delaying deployment. These aren’t product problems. They’re cost problems. For the past two
By Alan Jacobson, AI Economics Strategist Meta is reportedly planning layoffs of up to 20% of its workforce while simultaneously ramping AI infrastructure spending. The assumption is straightforward: fewer employees, more AI, higher efficiency, better margins. That assumption may be wrong. AI changes the cost structure of the business. Labor
By Alan Jacobson, AI Economics Strategist According to The Information on Wednesday, March 18, 2026: OpenAI has decided the “everything everywhere all at once” strategy isn’t ideal, after all. OpenAI executive Fidji Simo told staffers last week that the company needed to “refocus on business customers and cut down
By Alan Jacobson, Systems Architect & Analyst AI is not ready for business. That isn’t an opinion. It’s what the data, the vendors and real-world trials now show: * MIT researchers report that roughly 95 percent of generative AI pilots inside companies fail to move beyond experimentation. * A recent
Despite endless comparisons to Amazon, Google, and Facebook, AI is not like monetizing eCommerce, Search, or Social – because unit economics improve with scale for them. But not for AI. Rubber duckies are cheap to make one at a time, and cheaper to make a million at a time. Amazon loses