How Big Is the Instructor-Led Training Market? About $30 Billion a Year in the U.S.

The U.S. instructor-led training market is worth approximately $30 billion per year, with a defensible range of $25 billion to $35 billion. That figure covers training sold commercially and delivered by a live human instructor, whether in a classroom, on a customer site, or in a virtual classroom. It excludes internal L&D delivered by a company's own trainers on payroll, and it excludes self-paced e-learning.

I built this number because nobody publishes it. Every source you can find measures something adjacent: total corporate training spend, the global e-learning market, or a single vertical. None of them answer the question a training provider actually asks, which is how much money changes hands in the U.S. every year for training that a person teaches.

Key findings

• U.S. commercial instructor-led training runs about $30B a year. Three independent methods converge on $25B to $35B.
• Strip out pre-employment vocational tuition and it is $22B to $27B. Trade school and CDL tuition is instructor-led and commercially sold, but it is career-entry education rather than workforce training. Decide which definition you are using before you quote a number.
• The widely cited $102.8B "training market" figure is not this market. That is total U.S. corporate training spend at organizations with 100 or more employees, and $64.7B of it is internal training staff payroll. Only $16B is external purchasing.
• Instructor-led delivery is not dying. It is sorting. Where a credential can be satisfied by seat time alone, self-paced online has taken 60% to 90% of volume and collapsed the price. Where a human has to watch you perform a skill, live delivery has lost essentially nothing.
• OSHA 10 and 30-hour outreach training went from 70.2% classroom in FY2016 to 39.2% in FY2025 (OSHA). The fastest documented modality shift in the dataset, and it happened in nine years.
• The virtual classroom replaced the physical one, not the instructor. 24% of U.S. corporate training hours are now delivered virtually with a live instructor, against 28% in a physical classroom.

What counts as instructor-led training

Scope discipline is where most market-size claims fall apart, so here is mine.

In scope.Training sold at arm's length by a provider to a learner, an employer, or a sponsor, where delivery is live and human-facilitated. That includes in-person classroom, hands-on labs, on-site delivery, and synchronous virtual classroom (VILT). It covers commercial training providers, vendor product training sold directly or through authorized partners, association-delivered education, university non-credit and executive education, licensure and continuing education, apprenticeship instruction, and skilled trades.

Out of scope.Internal L&D delivered by a company's own trainers to its own employees. Self-paced e-learning, content libraries, and subscription catalogs. Degree-granting higher education. K-12. Consumer and hobby instruction.

Subject matter is irrelevant to the definition. Sales training, leadership, cybersecurity, welding, nursing CE, forklift certification, Kubernetes, and CPA continuing education all count the same way, because the criterion is delivery mode and commercial sale.

Why this number does not already exist

The NAICS codes do not match the industry.Census tracks "Professional and Management Development Training" (611430) and "Computer Training" (611420), but an enormous share of commercial training is sold by companies classified as something else entirely: consulting firms, professional associations, software vendors, hospital systems, universities, and roughly 900,000 nonemployer sole proprietors who teach for a living.

The analysts measure spend, not delivery. Training Magazine and Training Industry Inc. both size total training spend including internal payroll. Neither breaks out how much of external purchasing is live instruction versus content and technology.

Nobody discloses.I checked AWS, Microsoft, Salesforce, Cisco, Oracle, SAP, ServiceNow, Databricks, NVIDIA, Google Cloud, Adobe, and Autodesk. Not one reports instructor-led training revenue separately. It disappears into "Services" or "Professional services and other." On the industrial side, Siemens, Rockwell, Fanuc, Caterpillar dealer technical training, and Epic Systems disclose nothing at all.

So I triangulated. Three methods, built independently.

Method 1: Census bottom-up

The 2022 Economic Census is the only official source with six-digit NAICS revenue. These are firms whose primary business is training.

Grow that to 2026 at roughly 4% a year and you get about $39B. Take the live-instructor share, around 60% for this mix, and you have $23B. Then add the training sold by firms classified under other NAICS codes: associations, consultancies, technology vendors, university non-credit programs, hospital systems, and independent contract instructors. That is another 30% to 40% of the total, bringing the Census route to $33B to $38B.

One detail worth knowing: apprenticeship training is 93% tax-exempt. There are 84 for-profit apprenticeship training firms in the entire country. Nearly all of it runs through union JATCs and employer-funded programs.

Method 2: Analyst top-down

Training Industry Inc. puts North American total training spend at $186.4B for 2025, with approximately 29% flowing to external suppliers. That is about $54B in North America, or roughly $47B in the U.S. Apply the live-instructor share of external spend, which is around half, and you get $23B. Their definition explicitly excludes individual-paid continuing education and vocational tuition. Add those back and the analyst route lands at $28B to $32B.

Method 3: Segment build-up

This is the method I trust most, because it forces you to name where the money is.

Three methods. Three different starting points. All three land between $25B and $35B.

The $102.8 billion mistake

If you have ever sat in a board meeting about corporate training, you have seen the number $102.8 billion. It comes from Training Magazine's 2025 Training Industry Report, and it gets quoted as though it were the size of the training market. It is not. Inside it: $64.7B in internal training staff payroll, $22.1B in travel, facilities, equipment and admin, and $16.0B in outside products and services.

Nearly two-thirds of that headline number is the salaries of people who already work at the company. Only $16B leaves the building. And the survey universe is U.S. organizations with 100 or more employees, so it excludes government, small business, individual-paid continuing education, licensure, trades, and vendor product training sold to customers.

The instructor-led slice of that specific number is $16B multiplied by roughly 50% live delivery, or about $8B. That is a real figure. It describes corporate L&D budgets only. It is one component of the $30B market, not a competitor to it.

The takeaway: when someone cites $102.8B at you, they are quoting a payroll line. Ask which part of it they think you can sell into.

What is growing and what is dying

This is the part that matters more than the total, and it is the finding I did not expect. Instructor-led training is not in general decline. It is sorting into two piles, and the rule is brutally simple: can a human verify the skill without watching you do it?

Wherever a credential can be satisfied by seat time alone, self-paced online has taken 60% to 90% of the volume and crushed the price to a few dollars per credit hour. Wherever a psychomotor skill has to be demonstrated to a human being, live delivery has kept essentially all of it.

The second pattern is just as important, and it gets lost constantly. In corporate leadership and sales training, the shift over the past six years was not from live to self-paced. It was from in-person to virtual instructor-led. Twenty-four percent of corporate training hours now happen in a virtual classroom, nearly matching the 28% delivered in a physical one.

The instructor did not go away. The room did.

What this means if you run a training business

Your defensibility is what your instructors do that software cannot. If your catalog is mostly knowledge transfer a learner could get from a recorded video and a quiz, you are on the wrong side of the sorting line, and the OSHA numbers tell you how fast it moves once it starts. If your delivery involves hands-on practice, live assessment, real equipment, simulation, or a skills check somebody has to sign off on, the substitution risk has largely already played out around you. Most providers have both. The exercise worth doing this quarter is running your catalog against that line and seeing what share of revenue sits on each side.

The shift to VILT changed your operating model, not your market. Virtual delivery did not shrink demand. It shattered the constraints that used to make scheduling simple. When every class was in a room in a city, geography did your instructor matching for you. Now your instructor pool is national, your delivery windows span time zones, and a single client wants the same course delivered dedicated, open-enrollment, on-site, and virtual inside the same quarter. That is why training operations became the bottleneck. The demand side got easier. The coordination side got much harder.

Growth here comes from margin, not just bookings. The commercial instructor-led training market grows at low single digits. The providers who grow meaningfully are not the ones who simply sell more. They are the ones who deliver more without adding proportional coordination headcount. My co-founder and I grew an instructor-led training company to the point that Pluralsight acquired it and made it their instructor-led arm. What capped us was never demand. It was the number of coordinators, planners, and schedulers it took to turn a booking into a delivered class. Every new region, modality, and client requirement added another person and another spreadsheet.

That is the real constraint on this $30 billion market, and it is an operations problem rather than a demand problem.

Methodology and limits

I used three independent methods and report the convergence rather than a single point estimate. Where a figure is derived rather than reported, I say so.

Hard sources. U.S. Census Bureau 2022 Economic Census (table EC2261BASIC) for six-digit NAICS revenue. OSHA Outreach Training Program growth data for the classroom and online split by fiscal year. ACCME annual data reports for accredited continuing medical education income. Department of Labor registered apprenticeship counts. FMCSA Entry-Level Driver Training Provider Registry. SEC filings and IRS Form 990 program service revenue.

Derived estimates. Medical device and clinical skills training, higher education non-credit revenue, and privately held sales and leadership providers are derived, not reported. Higher education non-credit is the weakest link: NCES states plainly that no national data collection captures non-credit enrollment or revenue, so treat that $2.6B as a placeholder rather than a finding.

Known limits. The largest swing factor is whether you include pre-employment vocational tuition, which moves the total by $5B to $7B. Nonemployer sole-proprietor instructor revenue is visible only at the sector level, so its allocation is estimated. Private training companies do not disclose revenue, so the sales and leadership segment holds up because top-down and bottom-up agree, not because any individual figure is verified.

Using this research? Please cite it and link back to this page. If you have data that would sharpen any segment, especially higher education non-credit or industrial OEM training, I want to hear from you.