The Summer Class Forecast

"We learn wisdom from failure much more than from success." — Samuel Smiles

Summer in New York brings the return of many beloved seasonal favorites: Bryant Park Movie Nights, Aperol spritzes on rooftop bars, and the sudden appearance of summer associates in BigLaw.

If the flock of 1Ls and 2Ls in your office looks a little thinner this year, you are not imagining it. Despite a broader hiring uptick across law firms, summer associate and first-year placements remain below their pre-COVID levels. At the largest firms, summer classes have reportedly shrunk to their smallest size since 2012. Interestingly, law school applicants are up sharply, with some counts placing last year's increase as high as 33%, and another record-breaking applicant cycle expected this fall.

On the surface, both trends are explainable. Law schools have historically seen higher traffic during periods of economic uncertainty and political unrest. BigLaw, meanwhile, experienced a COVID-era hiring boom, and demand has cooled since.

Still, it would be easy to miss the shrinking junior population beneath the headlines advertising growth in BigLaw. One recent headline even referred to the growing DFW market as "Y'all Street." There is demand in BigLaw for lawyers, but that demand does not appear to be concentrated at the bottom of the pyramid.

A premium has been placed on mid-level and senior associates, and not only by the firms they once scrambled to join as freshly barred juniors. Legal AI companies are chasing them, too. Who better to sell legal technology back to law firms than the people who came out of them?

Much as I love to speculate on lawyer business, this is still a legal ops blog, and I would not be true to it if the story stopped at lateral movement and hiring charts. These numbers sat on my desk this week: lighter summer classes, smaller first-year cohorts, more mid-level and senior associates lateraling in like gifts from the Law Stork (patent pending). Eventually, I started to wonder what all of this means for the future of legal work, and more specifically, how we price it.

The obvious read is that BigLaw is cutting back on junior hiring because junior work is the work most exposed to AI. While reductionist, there is probably a significant truth in that take. Much of what once kept first-years reliably busy is now easier to compress, delegate, or automate.

The less obvious read is that firms may also be stepping away from a training model the market is no longer willing to subsidize quite so generously.

For years, the core economics of matter management worked because junior labor was inexpensive enough to absorb inefficiency. The leverage model allowed firms to train lawyers on the client's dime, though rarely in those words. I remember a particularly direct response from a partner when a client pushed to make first-years non-billable. She agreed easily, then told us that if they preferred to pay more for a senior lawyer to do the legwork of a junior, she would happily oblige.

There was a certain logic to that answer, at the time. Now, AI compression makes the argument harder to defend, and the practice behind it harder to execute.

The leverage model served another function, too. It trained and cultivated talent in-house. Enthusiasm may be a native trait, but real skill in any industry takes time to grow. I asked a senior associate what she remembered from her junior years, and she recalled poring over hundreds of pages of disclosure schedules. It was tedious, lengthy work, but those grueling hours, she said, taught her where to spot the anomalies in a way AI cannot yet reliably manage.

Human judgment has yet to be commoditized. However, if the divide continues to grow between the value of labor and the value of discernment, the question becomes whether the current model can produce enough judgment-capable lawyers to sustain itself.

I think the answer may come down to whether firms use AI to aid learning or simply accelerate production. The distinction is captured less by what the tool can do than by how it is used.

If a junior lawyer uses AI to get to a first pass faster, and then is encouraged (by either the model or the firm) to engage meaningfully with the model: input clarity, output testing, checking the sources, weighing the judgment calls, and ultimately being able to defend the result, there is still a learning loop. The opportunity may lie not only in efficiency, but in designing tools and workflows that invite that kind of engagement rather than bypass it.

Conversely, when the tool becomes a shortcut that also results in intellectual atrophy, the lawyer may get faster without getting sharper.

The premium exists in knowing when to question the first answer, test its reasoning, and recognize when something important has been missed.

Firms reducing investment in junior talent may also be responding rationally to a market that no longer wants to pay for apprenticeship disguised as leverage. Rack rates are climbing, and clients are asking increasingly consequential questions about staffing mix, predictability, and whether they are paying for effort or value.

That leaves firms with a difficult strategic tradeoff: invest in training now and seed the judgment premium for the future, or capture the maximum efficiency gains on the front end and hope there are still battle-tested mid-levels available in four to five years.

So, knowing this, should we price judgment, and how so?

I have seen, and have written about, the idea that the billable hour is dying because clients are demanding a more value-forward price for legal work. But as we have established, the billable hour is not merely a client-facing vanity. It is a load-bearing institutional feature.

Maybe the pricing question is not really "billable hour vs not." Maybe the better question is how to distinguish between hours that are valuable because they reflect judgment, and hours that are becoming commoditized because technology can compress them.

A pricing model that acknowledges both judgment inflection points and efficiency compression would signal something important. It would show that the firm understands the work is changing, and that not every hour carries the same kind of value.

For spend-conscious clients, matter management may become even more important in winning pitches. If the talent pyramid is changing, buyers have a right to ask how the work is being staffed, where the judgment sits, and what the firm is doing to maintain quality without hiding cost inside leverage.

Being able to answer those questions is not merely a response to procurement pressure. It is good legal ops.

The two speculative fears I hear most often are the invasion of RoboLawyers (patent also pending) and the idea that AI will simply make lawyers worse. I think one is unlikely, and the other is too certain for a question this intricate.

The more important issue is whether firms will design AI workflows that preserve the intellectually demanding aspects of professional development: the friction that turns exposure into judgment. That is a design choice, but it is also a pricing one. If firms get that wrong, the market may end up with more efficient output and less durable judgment. That outcome makes the pricing dilemma worse, not better.

The market is becoming increasingly adept at producing work. It is still grappling with how to value the people who discern what should be done with it.