[AI SPRINT] Everyone Is Asking the Wrong Question About AI and Jobs

You've likely seen Jack Dorsey's recent letter to Block shareholders. The second paragraph said "2025 was a strong year." The first paragraph announced 4,000 layoffs — nearly half the company's workforce.

"The core thesis is simple," Dorsey wrote. "Intelligence tools have changed what it means to build and run a company. I don't think we're early to this realization. I think most companies are late."

He predicted other companies would reach the same conclusion within the year. That prediction already looks less like provocation and more like pattern. This week, Reuters reported Meta is planning layoffs that could top 20% of its workforce. Atlassian cut 10% of its staff. The week before, the story was Block. The week before that, Amazon.

This is why the AI jobs conversation has entered a new phase. Not just because of the number of cuts, though those matter. Because of the reasoning behind them. Profitable companies with growing customer bases are cutting thousands of roles not because the business is failing, but because the work itself is changing.

That is the context for everything below. First, a quick note on how we can help:

What's Actually Happening with Jobs

AI and jobs is now one of the loudest debates in business. Most of the commentary is not useful.

The pessimists point to the headlines: layoffs, restructuring, automation, another week, another AI-attributed cut. The optimists reach for history: every major technology eventually creates more jobs than it destroys. Steam. Electricity. The internet. It always works out.

I've never been comfortable with that answer.

It may prove true again. But right now it is mostly a hope, not a plan. Three and a half years after ChatGPT launched, we can point to important new roles and new companies, but not yet to new categories of employment at the scale needed to offset broad displacement. Anyone telling people not to worry should have a more specific answer than "history says this usually works out."

The visible layoff numbers only tell part of the story. Of the 1.2 million job cuts announced in 2025, AI was cited in about 55,000 — roughly 4.5%. Significant, but not the full picture. In 2026 the pace has already accelerated, with tech layoffs running at 736 people per day, up from 674 per day across all of last year.

But the deeper shift is quieter. Companies are not always firing. Often, they are just not backfilling. Teams shrink through attrition. Junior hiring slows. Expectations rise for what one person can now produce. One workforce analyst put it sharply this week: "We are essentially burning the bottom rungs of the career ladder to heat the house. Productivity is up today, but the talent pipeline for 2035 is looking dangerously empty."

That is the slow version of what companies like Block are doing fast.

I have seen both versions up close. Last week, I led an AI strategy session for a large real estate group projecting a 30 to 40 percent job reduction over the next five years. Their honest assessment: nobody in the industry has been taking AI seriously enough. By the end of our session, they had collapsed a five-year plan into a two-year goal.

Later, I met with a wealth management group. Their initial comfort was that wealthy clients will always want a human advisor.

That is probably true for some clients. But that is not the whole market. Many more people will be satisfied using AI for services wealth managers once provided exclusively. The industry does not need to disappear for the economics to change. It can simply compress: fewer hours per client, lower fees, more volume required to maintain the same income. The human advisor survives. The industry shrinks around them.

If you have not seen that pressure clearly in your own sector yet, that does not mean it is not coming.

Why the Standard Reassurance Falls Short

The standard reassurance goes like this: every major technology wave creates new jobs and new industries. The automobile created factories, mechanics, highways, gas stations. The computer created software, semiconductors, and the internet economy. The market will adjust.

That may still prove true. But the comparison has real limits.

Those technologies made physical things cheaper to produce and move around. The new jobs they created were easy to see — factories, roads, supply chains. The work was concrete and obvious.

AI makes thinking cheaper. That changes the cost of planning, research, analysis, coordination, and decision-making. It does not automatically create a new layer of physical infrastructure that needs people to build and run it. It can remove jobs faster than it creates whole new categories of work.

That is why the standard reassurance feels insufficient right now. Not necessarily wrong. Just too vague, too early, and too detached from what firms are actually doing today.

If you want to understand your own exposure, AI researcher Andrej Karpathy recently published a tool scoring 342 U.S. occupations on their AI replacement risk. Worth looking up your own job at karpathy.ai/jobs. The pattern is clear: work that lives on a screen is far more exposed than work requiring physical presence in the world.

Inside major institutions, the shift is already real and deep. This week I spoke with someone who worked in AI at a major financial services company. In the last 12 months alone, they built over 10,000 AI agents across the firm. That is not a pilot. That is a fundamental change in how work gets done inside one of the largest employers in financial services, while many competitors are still writing strategy documents.

So far, AI development and implementation roles do exist and matter. But they employ far fewer people than the broad base of knowledge work now being affected.

CNBC's Steve Sedgwick asked the question directly after the Block announcement: "I keep getting told on CNBC that AI will create new jobs to replace those being lost. I've been asking the same question for years. What are those jobs?"

The honest answer right now is: not enough, and not fast enough.

But there is a more useful frame than waiting for new industries to appear.

The Real Opportunity: Demand Creation

This is the part too few people are talking about.

AI does not just cut costs. It makes services possible that simply could not exist before — not because they were technologically out of reach, but because they were too expensive to deliver at a price people could afford.

That is the real opportunity.

Not necessarily brand-new industries. New demand. New markets. New service models. Work that could have existed before, but could not be delivered profitably at the right price, by the right person, at the right scale.

For years, many services were held back by the cost of expertise. The research took too long. The preparation was too involved. The work required too much support around one person to make the numbers work. That pushed these services out of reach for smaller businesses, thinner markets, and local communities.

That is changing.

Consider the fractional CFO. A single person can now offer monthly financial strategy, cash flow forecasting, and operating reviews to 20 or 30 small local businesses. Previously, that service required a full firm and large clients to be worth the effort. The demand was always there — small businesses have always needed this kind of support. What was missing was a cost structure that could reach them. AI reduces the research and preparation time enough that one capable person can now serve that market profitably.

The same logic applies across many fields. Local compliance and regulatory support — zoning, permits, labor law, environmental requirements — has always been underserved outside major cities because the traditional model was too expensive to bring to smaller clients. Healthcare navigation, where patients have always needed help interpreting bills and making sense of a fragmented system, but the time involved made it impossible to price in a way most people could afford. Real estate, where some transaction roles will shrink, but new specialized services around the physical and relational side of the process may emerge that never existed as a viable standalone offering before.

And then there is live events. Last month I worked with a global event production company on their AI adoption strategy. High-quality live events have always required professional sound, lighting, logistics, and promotion budgets that put them out of reach for most local organizations. AI is changing the cost of putting those events together. What if a local business district could now afford a proper concert series? Local musicians get a stage. Local businesses get foot traffic. Communities that have been slowly losing their gathering places get them back. The demand for connection was always there. The economics finally work.

The pattern is consistent: the demand existed, the economics did not, and AI closes that gap.

The question every leader and entrepreneur should be sitting with: what service could now exist because intelligence is no longer the expensive part?

What This Means Right Now

The jobs debate will continue. More companies will cut. Economists will keep arguing about timing and scale.

But for leaders, operators, and ambitious professionals, the practical questions are already clear.

1. Do not wait for macro certainty. The companies pulling ahead are not waiting for the debate to resolve. Every quarter you delay serious AI adoption increases the gap between you and competitors who are redesigning work now.

2. Look beyond cost reduction. Cutting cost is the obvious use case and the most crowded one. The more interesting question is where cheaper intelligence lets you serve customers you could not profitably reach before.

3. Watch for compression before collapse. Most industries will not vanish. They will compress: fewer hours per client, lower prices, higher expectations, more output from smaller teams. That shift is easier to miss and harder to recover from.

4. Think like an entrepreneur, even inside an existing business. The key question is not just "how do we defend what we have?" It is "what can we now offer that was not possible six months ago?"

The disruption is real. The labor impact is not theoretical anymore. And the reassurance that new jobs will appear is not sufficient on its own.

But there is a more useful lens. AI is not only a force for efficiency. It is a force for demand creation. It can unlock services, markets, and customers that were previously out of reach.

That is where a lot of the next wave of opportunity will come from.

What Happened in AI This Week

Four things worth knowing:

• OpenAI released GPT-5.4 mini and nano — smaller, faster, cheaper versions of their flagship model built for high-volume work. Mini runs more than twice as fast at a fraction of the cost. The price of running AI at scale just dropped again.

• ChatGPT can now take actions inside Google and Microsoft tools — drafting emails, creating documents, scheduling meetings — directly through the chat interface. AI embedded in the tools your team already uses, taking action on your behalf, is a meaningfully different category than AI you visit in a separate tab. Might require configuration by your systems admins to enable.

• Anthropic launched a formal certification program for Claude this week, alongside a $100 million partner network. Accenture is training 30,000 professionals on Claude. Cognizant is training up to 350,000. AI expertise is professionalizing fast, and the gap between organizations building that capability and those not is now measurable.

• Perplexity held its first developer conference and announced Computer for Enterprise — a multi-model AI agent with Slack integration and direct connections to business data, aimed squarely at replacing Microsoft Copilot and similar tools. Their bet: enterprises want the best AI model routed automatically to each task, not a single vendor's answer for everything.

Trent Gillespie is CEO of Stellis AI and a keynote speaker helping business leaders understand and operationalize AI in their companies. He spent almost nine years leading global innovation efforts at Amazon before leaving to help other companies build the capabilities they need to compete. Book Trent to speak to your group or book a call to discuss using AI within your business.

Did someone forward this newsletter to you? If you're not already signed up, you can subscribe to AI SPRINT™ for free here.