In September 2026, legal AI company Harvey raised $550 million in a single funding round at a valuation of $15.5 billion.
That number tells us where investors think legal services are going. The more interesting question is whether law firms are ready to follow.
AI is already compressing the time required for legal research, contract review, discovery, due diligence, compliance, and routine analysis. Corporate legal departments are adopting these tools quickly, and clients increasingly expect the efficiency to show up in the service they receive and the price they pay.
Yet the legal industry still depends heavily on a model built around time.
That creates a fundamental conflict.
AI removes hours from work. Law firms have traditionally sold hours.
The billable hour rewards something AI is designed to eliminate: time.
AI Is Pulling Legal Expertise Apart
Legal services have traditionally bundled many different activities together.
Research. Document review. Drafting. Analysis. Negotiation. Strategy. Judgment.
The client bought the bundle, and much of that bundle was priced according to the number of human hours required to produce it.
AI pulls those activities apart.
Knowledge is different from judgment. Research is different from strategy. Drafting is different from persuasion. Process is different from expertise.
I think of this as the expertise unbundling.
A client once had little choice but to purchase all of these capabilities together. AI makes some parts of that bundle increasingly abundant and inexpensive.
That exposes an uncomfortable reality about legal pricing.
Some work commanded premium prices because it required a large amount of expert human time to produce. The underlying task itself did not always require premium judgment.
As AI reduces the time involved, clients can see that distinction much more clearly.
This same economic pressure appears in other AI business models. In Multi Agent AI and the End of the Per Seat Model, I explored what happens when technology changes the unit of production while companies continue charging according to the old one.
Legal services are approaching the same moment.
Great Lawyers Become More Valuable
AI can make research faster. It can review enormous document sets. It can produce drafts, summarize cases and surface patterns across contracts.
As those capabilities become easier to obtain, the value of the remaining human contribution becomes much more visible.
Judgment. Negotiation. Persuasion. Creativity. Strategy. Relationships. Accountability.
Those are the capabilities clients wanted from exceptional lawyers in the first place.
Routine cognition becoming cheaper can actually increase the value gap between average and extraordinary lawyers.
A mediocre lawyer who built an invoice around hours of process faces obvious pressure.
An extraordinary lawyer whose judgment prevents a $50 million mistake becomes easier to value because technology strips away much of the process surrounding that judgment.
AI will make average legal work cheap and exceptional legal judgment priceless.
That is why I do not expect AI to commoditize the entire profession. It can commoditize significant parts of legal production while putting a greater premium on the capabilities that depend on experience, context, and judgment.
The $800 Hour and the $100,000 Decision
Imagine two lawyers facing the same problem.
One uses AI together with years of experience and reaches the right decision in twenty minutes.
The other takes twelve hours to reach the same conclusion.
Under a pure hourly model, the slower lawyer generates the larger invoice.
That becomes increasingly difficult to defend once clients know the same tools exist.
At the same time, pricing the first lawyer’s decision as twenty minutes of work makes little sense either.
If that decision saves the client $10 million, its value has almost nothing to do with the time required to reach it.
AI destroys the $800 hour and creates the $100,000 decision.
That does not mean every legal matter suddenly moves to value-based pricing. Hourly billing will continue to make sense in many situations.
It does mean firms will increasingly have to explain what the client is actually paying for.
Time becomes a weaker proxy for value once technology can radically change how much time a task requires.
What Is Really Slowing Legal AI Adoption
There are obvious barriers to AI adoption in law.
Confidentiality matters. Hallucinations matter. Regulation, professional responsibility, and liability matter.
Those risks are real.
There is also an economic problem sitting underneath them.
Why would a partnership aggressively deploy technology that reduces the number of billable hours required to produce the work it sells?
Law firms have a business model that can punish them for becoming more productive.
That conflict helps explain why AI adoption cannot be reduced to buying the right software. Firms also have to decide what happens to their economics when the software works.
Clients are likely to force that decision.
Corporate legal departments are already using AI internally. Deloitte’s 2026 research found that 61% of legal departments had moved into AI deployment, while 78% of legal leaders said cost reduction was the leading benefit they expected from external providers using AI.
Once one firm can deliver comparable or better work dramatically faster, the client develops a new expectation.
Expectations rarely move backward.
The greatest threat to a law firm is the firm across the street that changes the client’s expectations first.
Buying AI Is the Easy Part
A law firm can purchase the same AI platform as its competitors tomorrow.
That does not mean it has transformed anything.
The harder work is redesigning workflows, deciding where human review actually matters, changing pricing where appropriate, and making someone responsible for how humans and AI work together.
A tool sitting beside the old process simply adds another layer to the old process.
The quality of human oversight matters too. I explored this problem in The Human in the Loop Is Legal Fiction. A workflow does not become safe or accountable because a policy says a human reviews the output. That person needs the expertise, authority, and time to perform a meaningful review.
The firms that move fastest will be willing to cannibalize parts of their existing economics before competitors force them to.
That requires more than technology adoption. It requires operating model redesign.
Who Trains the Next Generation of Great Lawyers?
There is another problem that may take longer to appear.
Junior lawyers traditionally developed expertise through the same work AI is increasingly able to absorb.
Research. Document review. Drafting. Repeated exposure to transactions, arguments, cases, and mistakes.
Those tasks produced client work, but they also produced senior lawyers.
If AI removes a substantial amount of that work, firms gain immediate productivity while potentially weakening the system that develops future expertise.
This problem extends far beyond law. Automation often removes work that appears low value while quietly serving a high value developmental purpose.
I discussed the broader workforce version of this in AI Didn’t Take Your Job. Someone Using AI Did.. In law, the question becomes particularly important because judgment grows through repeated exposure to real problems.
If AI does the work that taught junior lawyers how to become senior lawyers, who trains the next generation of great lawyers?
Firms will have to create those learning experiences intentionally.
For decades, thousands of billable hours doubled as professional development. Once AI absorbs part of that workload, apprenticeship can no longer remain an accidental byproduct of the billing model.
What the Winning Law Firm Looks Like
Three years from now, the strongest AI-enabled law firms will have done more than add AI to existing workflows.
They will have redesigned what clients are actually purchasing.
Routine cognition will increasingly be automated. Human lawyers will concentrate more heavily on judgment, negotiation, persuasion, relationships, strategy, and accountability.
Pricing will gradually move closer to value, outcomes, complexity, risk and access to exceptional judgment. Hourly billing will remain part of the model, but firms will become less dependent on it.
Someone inside the organization will also need to coordinate the growing relationship between human expertise, AI systems and business processes. That need resembles the AI orchestrator role emerging in other industries, where the real work is deciding how different forms of intelligence should operate together.
The most important shift, however, is the question firms ask themselves.
“How can AI help our lawyers do today’s work faster?” keeps the existing model intact.
I would ask something more fundamental.
If we were inventing legal services today and the billable hour had never existed, what would we sell?
Process Gets Cheaper. Judgment Gets More Valuable.
AI is exposing something clients have suspected for years.
Some expensive legal work commanded high fees because delivering expertise was extraordinarily labor-intensive.
AI changes that process.
A firm that continues charging exactly as though the process has not changed gives clients a reason to question the entire value equation.
Simply cutting prices creates a different problem. Exceptional human judgment should not become a commodity just because the process surrounding it became more efficient.
The opportunity is to make process cheaper while making judgment more valuable.
Stop charging premium prices for process. Start charging premium prices for judgment.
That is where I think the economics of legal AI eventually lead.
The firms that understand it early will use AI to expose the value of their best lawyers.
The firms that continue selling time will increasingly have to explain why the client should keep buying it.
