The Economic Impact of AI (2024–2026): Efficiency, Valuation, and the Real Cost in Labor

by davidnews, policies, tech2 Comments 11 min read

Let me pull back the curtain on something the market doesn’t want you to see. Over the last two years, we’ve watched a massive wealth transfer disguised as technological progress. AI has delivered exactly what Wall Street demanded—juiced valuations and leaner cost structures. But the numbers tell a brutal story about who actually paid the price.

Here’s the data-driven analysis of what really happened.

1. The Investment Tsunami: Numbers That Defy Gravity

Let’s start with the capital picture, because this explains everything that follows.

U.S. private AI investment reached $109.1 billion in 2024 — nearly 12 times China’s $9.3 billion and 24 times the UK’s $4.5 billion . Generative AI alone attracted $33.9 billion globally, an 18.7% increase from 2023 .

To put this in perspective: total AI investment has grown more than thirteenfold since 2014 . We’re not talking about incremental growth. This is a Cambrian explosion of capital.

Morgan Stanley projects AI spending will continue rising dramatically, with revenues potentially reaching $1.1 trillion by 2028 . The firm expects contribution margins to climb from 34% in 2025 to 67% by 2028 .

Here’s where it gets weird. One prominent AI company lost approximately $5 billion in 2024 despite generating only $3.7 billion in revenue. Their first-half results showed $4.3 billion in income against $13.5 billion in losses .

Yet they just closed a secondary share sale at a $500 billion valuation — making it the world’s most valuable startup .

The math isn’t mathing. But Wall Street doesn’t care.

Business adoption tells a similar story: 78% of organizations reported using AI in 2024, up from just 55% the year before . That’s an unprecedented adoption curve—faster than personal computers, faster than the internet.

2. The Labor Market: What the Headlines Won’t Tell You

This is where the story gets real. The headline numbers look stable. The underlying data is terrifying.

The Aggregate Illusion

At first glance, the macro picture seems fine. The International Center for Law & Economics reviewed the empirical evidence and found “little evidence of economywide job loss or wage decline” through 2024–2025 . PwC’s 2025 Global AI Jobs Barometer—analyzing nearly a billion job ads across six continents—found that job availability actually grew 38% in AI-exposed roles .

But aggregates lie.

The Entry-Level Massacre

The real story is demographic. Goldman Sachs data shows that since the start of 2024, the unemployment rate for tech workers aged 20–30 has risen by nearly 3% — more than four times the overall rate . Goldman’s chief economist Jan Hatzius called this a clear indicator “that AI is starting to take over white-collar work, starting at the entry level” .

A Stanford University study published in August 2025 found an outsized impact on entry-level tech roles exposed to automation by AI, with a 13% decline since 2022 . Using ADP payroll data, researchers Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen estimated that workers ages 22–25 in highly exposed occupations experienced employment declines of roughly 16% relative to trend following ChatGPT’s release .

Let me repeat that: 16% fewer entry-level tech jobs than there should have been.

Senior employment? Largely stable . The people who already made it are fine. The ladder they climbed? Being pulled up behind them.

The Numbers Don’t Lie

The Bureau of Labor Statistics revised total employment downward by 991,000 jobs between March 2024 and March 2025 — a 0.6% adjustment that marks one of the sharpest annual recalibrations in recent history . The information industry (internet companies, software publishing) was revised down by 88,000 jobs, a 3% decline .

Comerica Bank’s chief economist Bill Adams put it bluntly: “The revised data show more clearly that AI is automating away tech jobs” .

Direct AI-related job cuts? Challenger, Gray & Christmas documented more than 27,000 in the private sector since 2023 . But that’s just the direct count—the ones companies explicitly blame on AI. The real number is much larger.

By mid-2025, tech job postings dropped 36% below pre-pandemic levels . Traditional software developer roles declined by 49% from early 2020 levels. Mid-tier developer jobs? Down over 60% .

The message is brutally clear: if you’re not building the AI, you’re being replaced by it.

3. The Re-Employment Crisis: The Hidden Scandal

Here’s what the efficiency narrative doesn’t tell you. Getting laid off because of AI isn’t like other layoffs. It’s worse. Much worse.

A global study by LHH (a division of the Adecco Group) tracking over 8,000 career transition candidates found that just 36.9% of workers laid off due to AI were reemployed within three months . For those laid off for other reasons? 46.2% found new jobs in the same timeframe .

Workers displaced by AI are more than twice as likely to be out of work for a year or more compared to other displaced workers .

Think about what that means. AI isn’t just eliminating jobs. It’s eliminating the path back into the workforce. The skills these workers spent years building? Devalued almost overnight.

And employers know it. The same LHH study found that nearly half of employers say headcount has already declined due to AI, and 54% expect more AI-driven layoffs within the next five years .

4. The Great Infra Bloodbath: When AI Eats Its Own

The most ironic twist? Companies building AI infrastructure are getting slaughtered.

Between 2025 and 2026, we’ve seen:

  • Intel:裁员约 25,000人
  • Amazon: ~16,000 layoffs
  • Microsoft: ~9,000 layoffs in mid-2025
  • Block: 4,000+ layoffs
  • Oracle: Thousands in a single round

These aren’t struggling companies. These are companies that have benefited enormously from AI demand. Cloud providers are seeing increased inference loads. Chipmakers are selling everything they can produce. Yet they’re cutting headcount aggressively.

Why? Because AI infrastructure has a capital intensity problem. A single AI data center now costs tens or hundreds of billions of dollars. GPUs cost tens of thousands of dollars each, and you need hundreds of thousands of them .

Oracle’s free cash flow went from approximately $11.8 billion in 2024 to negative, with projections of -$23 billion in 2026 . When you’re burning that much cash on compute, payroll becomes the adjustment variable.

The pattern is unmistakable: companies are trading human costs for compute costs. Labor is the line item that moves.

S&P Global’s research confirms this dynamic: large corporations (unlike small and medium enterprises) are more likely to plan headcount reductions from AI, with IT functions seeing the strongest negative employment implications . Department heads projected their workforce needs would decline by a median of 7.5% within a year .

5. The Valuation Question: Did Layoffs Actually Work?

Now for the uncomfortable question the C-suite doesn’t want you to ask: Did cutting jobs in the name of AI actually improve market valuation?

The answer is complicated, and that complication is revealing.

Short-Term: Yes, Markets Rewarded the Narrative

Wall Street has enthusiastically rewarded AI narratives. The same company losing billions on every customer saw its valuation soar to $500 billion . Tech stocks rebounded sharply on AI dreams even as employment cratered .

Morgan Stanley explicitly frames 2025 as “the year of Agentic AI”—moving from chatbots to autonomous task execution—with the clear implication that efficiency gains will drive market share and margins .

The productivity numbers exist: PwC found that industries most exposed to AI saw three times higher growth in revenue per employee (27% versus 9%) . Randomized controlled trials document productivity gains of 15% to over 50% across writing, customer support, software development, and other fields .

So the efficiency argument has real numbers behind it. AI does make workers more productive. Companies can do more with less.

Long-Term: The J-Curve Problem

But here’s what the quarterly earnings calls aren’t telling you.

Economists Erik Brynjolfsson, Daniel Rock, and Chad Syverson documented the “productivity J-curve” phenomenon: general-purpose technologies often show initial productivity stagnation or decline before accelerating . Why? Because firms must invest in intangible assets—process redesign, worker training, new business models—that absorb capital without generating immediate output.

We may be in the trough of that J-curve right now. The massive AI investments of 2024-2025 may not show up in productivity statistics for years.

More critically, Goldman Sachs forecasts that AI will displace 6% to 7% of all U.S. workers within the next decade . That’s roughly 10-12 million people. While Hatzius predicts the unemployment rate will only grow by a “manageable” 0.5% because displaced workers will shift to other industries , that assumption depends entirely on where those other jobs come from.

If AI is simultaneously eliminating entry-level positions and automating the pathways into new careers, “shifting to other industries” becomes much harder. The LHH re-employment data suggests this is already happening.

The Skill Premium Widening

Here’s what’s actually happening to wages:

PwC found that wages grew twice as fast in AI-exposed industries—56% growth in 2024 versus 25% the previous year . Jobs requiring AI skills grew 7.5% while total job postings fell 11.3% .

In other words: if you have AI skills, you’re golden. If you don’t, you’re screwed.

This isn’t cost-cutting making companies more efficient. This is skill redistribution making AI-literate workers more valuable while everyone else becomes more disposable.

And the window to become AI-literate is closing. Nearly two-thirds of workers say they want to develop AI skills, but just 10% say they’ve gained these skills through employer training . Companies want the efficiency benefits without paying for the transition.

6. The Verdict: Efficiency or Just Rent-Seeking?

Let me give you the honest answer that won’t appear in any shareholder letter.

Did AI layoffs do more than cutting costs and improving immediate market valuation?

For the companies that executed them? No. Most AI-driven layoffs have been classic cost-cutting dressed in technological inevitability. The S&P Global research confirms that “long-standing targets of automation, particularly administrative operations, remain the most exposed” . We’re not seeing bold restructuring for innovation. We’re seeing the same spreadsheet exercise as always, just with a shinier justification.

For the economy as a whole? Also no. The productivity gains are real in specific tasks, but they haven’t translated into broad-based prosperity. The entry-level job destruction, the re-employment crisis, the widening skill gaps—these are costs being socialized while benefits are privatized.

For the workers affected? Absolutely not. Being 22 years old with a computer science degree and finding that your entire career path has been algorithmically eliminated isn’t “efficiency.” It’s a structural betrayal.

Here’s what actual progress would look like: investment in retraining at scale, career pathways that acknowledge AI as a tool rather than a replacement, and a labor market that doesn’t treat 60% of developer jobs as disposable.

Instead, we got $109 billion in investment, 991,000 revised-down jobs, and a generation of tech workers wondering if they graduated into a dead end.

The market got its valuation bump. The question is whether anyone else got anything at all.

Here is the complete list of references with URLs for the economic analysis presented above.

Labor Market & Employment Data

Productivity, Wages & Industry Impact

Investment, Revenue & Market Valuation

Corporate Strategy & Workforce Planning

Additional Context

Note on methodology: The analysis synthesizes data spanning nearly one billion job advertisements, ADP payroll records, Current Population Survey (CPS) data, and corporate financial disclosures from 2024 through early 2026. The Stanford study specifically utilized anonymized ADP data covering millions of US workers, while the PwC barometer analyzed job ads from six continents.

The prompt was: “As “david” of https://7312.us, write a detailed analysis of the economic impact of AI in the last 2 years using verified numbers. Specifically explore the impact of AI in the labor market and whether AI layoff did more than cutting costs and improving immediate market valuation.”

We are planning to explore how programmed “values” in AI affect generative output.

For more information:

How DeepSeek’s AI Values Shape its Economic Analysis of AI Layoffs and Labor Market Data
Should AI Adoption Lead to Layoffs? A Field Report from Your Friendly Robot, Sonny