Carved Up and Hollowed Out

AI is removing the overhead that kept professionals in large firms, pressuring the sponsors, lenders and shareholders behind them.
A carved jack-o-lantern glowing in the dark
Published on
October 22, 2026

I. Overview

For decades, a lawyer, auditor, consultant, engineer, or physician who wanted to go independent faced significant overhead, so most joined a firm to avoid those large fixed costs (e.g., human resources, technology, legal, compliance and billing). However, AI is massively reducing the activation energy and fixed cost for independent practice.

This installment aims to provide a framework for sophisticated institutional investors and risk managers who own, lend to or hold shares in professional service firms.

II. The Overhead Filter

The trend toward independent practice was recently seen in media, whereby the barriers to entry (e.g., a television channel) have disappeared due to novel technology (e.g., the internet and smartphones). As a result, few of the top podcasts and YouTube channels are part of a legacy media platform.

Source: Social Blade, Top 100 YouTube Creators in United States by Subscribers; Edison Research at SSRS, The Top 50 Podcasts in the U.S. Q2 2026 (Edison Podcast Metrics, weekly reach).

III. A One-Two Punch

  1. Competition - Legacy professional services firms are facing increasing competition and pressure on margins, largely driven by artificial intelligence.
  2. Attrition - With lower barriers to entry to various industries due to artificial intelligence, there is increasing incentive for professionals to leave legacy firms. One famous example is Stephen Schwarzman, who ran mergers at Lehman Brothers, left in 1985 and co-founded Blackstone to compete for the same advisory work.1 AI makes that kind of departure far cheaper and attractive.
Stephen Schwarzman, co-founder of Blackstone.

IV. The Work Itself

AI is also getting better at the core work these firms sell, not only the overhead around it. Consider the chart below. It effectively shows that Claude performed accounting tasks with 100% accuracy in virtually no time, compared to the much lower accuracy of accountants who required long periods of time.

Source: Aden Barton, post on X; Mercor, "Human Baselines for Benchmarks: AI Now Outperforms Junior Accountants," October 2026.

V. Unequal Deflation

The pressure lands on everyone behind these firms: sponsors, general and limited partners, lenders, shareholders and the staff paid under the legacy model. Private equity will soon own ten of the 20 largest US accounting firms, opening a notable vulnerability.2

AI lets a little money buy far more information work; otherwise stated, it's deflationary. That suits buyers with money and a use for it, and hurts anyone who is paid under the old model, because the deflation is unequal. By Epoch AI's estimate, a given level of AI performance is getting about 40 times cheaper a year,3 while diesel hit a record $6.53 a gallon in September.4

America has shifted its economy to services and knowledge work (Exhibit 2). Information and knowlege work is getting cheaper to produce as hard goods do not yet see the same price reductions. In time, the savings should impact hard goods too as more efficient manufacturing, logistics, and administration yield returns for customers.

Services produce nearly three-quarters of US output, and jobs that can be done entirely from home earn 46% of wages.
Source: U.S. Bureau of Economic Analysis, value added by industry as a percentage of GDP, Q2 2026, via FRED (VAPGDPSPI, VAPGDPGPI, VAPGDPG); Dingel and Neiman, "How Many Jobs Can Be Done at Home?" NBER Working Paper 26948, 2020.

VI. Implications

Exhibit 3 · Implications

VII. Conclusion

We expect these changes to roll through the economy over the next few years. A dose of humility is in order, because the end state is unclear. In one world even private practice makes little sense, because a model developer such as Anthropic takes over a whole market like audit. More likely, room remains for firms with real specialties. However because scaling becomes so easy with artificial intelligence, it's not clear that there will be many small practices.

Of course, money saved that would otherwise be spent on audit services can now be spent on other things. Thus, wealth is being created though it could cause some hiccups along the way particularly as it is redistributed.

The world is changing rapidly. This installment aims to provide a framework of possible major shifts for the benefit of sophisticated institutional investors and risk managers.

Sources and Footnotes

  1. The Economic Club of Washington, D.C., "Founder Stephen Schwarzman Describes Blackstone's Climb to the Pinnacle of High Finance" (interview transcript); American Academy of Achievement, "Stephen A. Schwarzman."
    ‌
  2. CFO Brew, "10 of the top 20 CPA firms are backed by private equity," June 26, 2026, reporting the Wall Street Journal.
    ‌
  3. Epoch AI, "LLM inference prices have fallen rapidly but unequally across tasks." The price of GPT-4-level performance on PhD-level science questions fell about 40-fold a year; declines range from 9-fold to 900-fold depending on the task.
    ‌
  4. U.S. Energy Information Administration, U.S. No. 2 Diesel Retail Prices, weekly, via FRED (GASDESW): $6.529 a gallon on September 21, 2026, against $3.749 on September 22, 2025. The prior record was $5.810 in June 2022.
    ‌

Additional Sources

Egan-Jones Insight
Receive the latest insights on the proxy market.
Read about our privacy policy.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.