Why AI Will Grow Some White Collar Jobs and Delete Others

On August 14, SpaceX closed its $60 billion acquisition of Cursor, the AI coding tool that reached roughly $2.6 billion in annualized revenue and counts more than half the Fortune 500 as customers. The interesting part is not the price. It is the mission SpaceX posted when the deal was first announced in April, which promised "the world's best coding and knowledge work AI." A product that spent three years learning to write software is now aimed at everyone who writes anything.

Around the same time, this viral post from SightBringer crossed my feed declaring the professional middle to be in slow liquidation, arguing that six figure salaries in the coordination layer were administrative overhead dressed up as expertise, and that the compression arrives within 12 to 18 months. I could not fully dismiss it, and I could not fully believe it either, because the labor data does not show a collapse while the Cursor deal shows exactly the ambition the collapse argument assumes. Working through that contradiction led me somewhere more useful than either camp.

The data shows a freeze and a slow bleed, not a liquidation

Start with what is measured. The New York Fed puts unemployment for recent college graduates at about 5.6 percent through the second quarter of 2026, with underemployment at 42 percent, meaning nearly half of new graduates are working jobs that never required their degree. The Economic Policy Institute's read of the JOLTS data shows the mechanism behind those numbers is a depressed hiring rate with quits falling alongside it, workers and employers sitting tight, while the layoff rate itself stays low. That is a freeze, and a freeze does not photograph well, which is why it produces neither crash headlines nor reassurance.

Underneath the freeze, though, there is a bleed. White collar sectors including finance, information, and professional and business services have been cutting jobs on net for three consecutive years while GDP stayed positive, a divergence that does not happen in normal cycles. Challenger, Gray and Christmas counted over 1.1 million layoff announcements in 2025, the highest total since the pandemic, and analysts now describe a pattern of continuous small reductions, frequently under 50 people at a time, that never trigger a headline. That invisibility is why I built the Job Market Blueprint, a daily, deduplicated layoff tracker mapping every event by state and industry alongside live hiring and unemployment data and what we are seeing in the data is that the doom posts are wrong about the present and the optimists are wrong about the trend. The honest description is a market that stopped growing the professional class and started quietly thinning it. (Also read: The Velocity Gap: Why AI Will Collapse Housing Before Washington Notices)

Coding was the test market, and the number of builders went up

Which brings the question back to Cursor, because coding is the one profession where this experiment has already run. Cursor made writing software dramatically cheaper, and the population of people building software expanded. The tool passed a million daily users, designers began shipping full-stack applications, and enterprise sales people with no computer science background, myself included, started building and publishing working products. (Also read: Vibe Coding Without a CS Degree)

There is a name for this and it is 161 years old. When James Watt's efficient steam engine cut the coal needed per unit of work, coal consumption exploded, because cheaper power made a thousand previously uneconomical uses viable. The economist William Stanley Jevons documented it in The Coal Question in 1865, warning that assuming efficiency leads to lower consumption gets the relationship backwards, because "the very contrary is the truth." Software demand behaved exactly like coal demand. When building got cheap, the world discovered it wanted far more software than it had ever budgeted for.

So if the coding precedent holds, Cursor's expansion into knowledge work should grow the knowledge workforce. That is the case for job gains, and it is a real case. It just applies unevenly, because demand for knowledge work splits into two kinds that respond to cheaper production in opposite ways.

The Elasticity Line decides which side of the tool you are on

Call it the Elasticity Line: the boundary between work whose demand expands when its cost falls and work whose demand is capped no matter how cheap it gets. Software sits on the elastic side, and so does any output the market keeps buying more of as the price drops, such as analysis that finds revenue, products that win customers, content that converts, and research that opens decisions. Make those cheaper and organizations consume more of them, which is why the coding population grew.

Coordination sits on the capped side. Status reports, meeting summaries, internal decks, and the administrative packaging of other people's work have demand set by organizational complexity, and no company wants more weekly updates because updates became free. The effect then compounds, because when AI thins a team, the coordination the remaining team requires shrinks with it. The tool cuts capped work twice, once by producing it faster and once by simplifying the organization that consumed it. This is why the same product can grow employment in one department while deleting the one next to it, and why the org chart, more than the model, determines who feels which effect. (Also read: Enterprise AI Is an Organizational Design Problem in Disguise)

The human cost concentrates on the capped side, among mid-career professionals who built fixed-cost lives (i.e. mortgages, tuition, and two-income budgets) on the assumption that a six figure coordination salary was durable. When the job goes, the mortgage does not, and a slow social downgrade grinds harder than a clean shock because it never presents a single moment to respond to.

Where does this leave the next year and a half?

So run the test on your own role. If everything you produced this quarter suddenly cost one tenth as much to produce, would your company want more of it or would it simply need fewer of you? An answer of more puts you on the elastic side, and the rational move there is to adopt these tools as fast as your company will allow, because your market is about to grow. An answer of fewer means the work sits on the capped side, where a salary was only ever as durable as the organizational complexity that justified it, and there is still time, though not unlimited time, to shift toward output that customers, revenue, or real decisions pull on. That shift is the difference between watching a tool like Cursor arrive in your department as a multiplier or as a replacement.

Key Takeaways

• The Elasticity Line separates work whose demand grows as it gets cheaper from work whose demand is capped by organizational complexity, and AI expands the first category while compressing the second.
• Current labor data shows a hiring freeze plus continuous small cuts, with white collar sectors shrinking on net for three straight years during positive GDP, rather than a mass layoff wave.
• Cursor made coding cheaper and the population of people building software grew, repeating the demand pattern Jevons documented with coal in 1865.
• Coordination work gets cut twice: AI produces it faster, and the thinner organization AI enables needs less of it.
• The practical test for any white collar role is whether the company would want more of your output if it cost one tenth as much to produce.

FAQ

What is the Elasticity Line?

The Elasticity Line is the boundary between knowledge work whose demand expands when its cost falls and knowledge work whose demand is capped regardless of price. Software, revenue-generating analysis, and customer-facing output sit on the elastic side. Coordination work such as status reporting, summarizing, and administrative packaging sits on the capped side, because its demand is set by organizational complexity rather than by a market.

Will Cursor's expansion into knowledge work cause job losses or job gains?

Both, sorted by the Elasticity Line. The coding precedent suggests agent tools grow the population doing elastic work, since cheaper building surfaced demand for far more software. The same tools compress coordination roles, because organizations do not buy more internal reporting when it gets cheaper and a thinner organization needs less coordination overall.

Is AI causing mass white collar layoffs right now?

The measured data says no. Layoff rates remain low, and economists at EPI and the New York Fed attribute most graduate labor market weakness to a hiring freeze rather than AI displacement. What the data does show is white collar sectors cutting jobs on net for three consecutive years during economic growth, through frequent small reductions rather than headline events.

About the Author

Martin Goetzinger has spent his career in enterprise software sales, helping large organizations such as Apple, Microsoft, and Verizon connect data, insight, and action. His work focuses on transforming how businesses measure success and create customer value through technology.

Outside the enterprise world, he writes about the five forces he believes are reshaping everything: AI, blockchain, energy, personalized health, and robotics. Not from a purely technical lens, but from a human one as to how these technologies will redefine work, wealth, and well-being.

He is based in the U.S. and publishes at www.MartinGoetzinger.com.

Disclaimer

The views expressed in this article are the personal opinions of the author and are provided for informational and educational purposes only. Nothing in this article constitutes investment advice, financial advice, legal advice, or any other form of professional advice. Do not make investment or financial decisions based on the content of this article. Always consult a qualified professional before making decisions that affect your finances, business, or livelihood.