It’s budget season, and the AI requests are already on your desk: more tool licenses, prompt-writing workshops, an AI learning platform. At the same time, the budget you’re dividing is getting smaller.
SHRM’s 2026 L&D benchmarking found the median organization spent $299 on L&D per full-time employee, down from $414 in 2025, even as training hours held steady at eight per employee. In a July 2026 Careerminds survey of 600 HR leaders and budget owners, 47% said their companies had raised AI spending while cutting L&D. And Deloitte’s analysis of its own clients’ AI spending, reported by Fortune, found 93% going to data, technology and infrastructure, with just 7% for the people side: redesigning work, training and managing change.
Smaller budget, bigger AI ask. So the question for 2027 isn’t whether to fund AI training. It’s what each dollar actually buys.
Two different things called “AI training”
Most AI budget requests mix together two investments that do different jobs.
Broad AI literacy reaches many employees: courses, tool training and prompt workshops. It raises everyday use and confidence. Every organization needs some of it, and many still have little. SHRM found that only about a third of organizations offered AI upskilling beyond basic compliance in 2026.
Deployment depth reaches a few people: the ones who can carry an AI initiative from pilot to production and own it afterward. That means scoping use cases, coordinating integration across vendors and IT, governing models, driving adoption and holding vendors to business results.
The trap is funding the first and expecting the second. Everyday AI use spreads largely on its own. MIT’s Project NANDA found workers at more than 90% of companies using personal AI tools for their jobs, even though only 40% of those companies had bought an official subscription. Deployment is what stalls: in the same research, just 5% of custom and vendor-sold enterprise AI tools reached production. We break down why AI pilots stall in a separate post. A prompt-writing workshop doesn’t fix any of the reasons.
A legacy lesson from L&D’s own playbook
Most L&D leaders know the 70-20-10 model. It traces back to Center for Creative Leadership research on how executives developed, summarized by Michael Lombardo and Robert Eichinger in 1996: roughly 70% of learning from challenging work, 20% from other people (mostly the boss) and 10% from courses and reading. As the Association for Talent Development has pointed out, the exact numbers have a thin evidence base. They come from executives looking back on what shaped them. But the model has lasted for decades because its core idea rings true: capability is mostly built on the job, with someone guiding you.
Deployment capability may be the clearest case of that idea. Nobody learns to carry an AI system into production from a course. They learn it on a live deployment, with a mentor, over time. A Registered Apprenticeship is essentially a structured, credentialed and partly funded version of the 70 and the 20. In the AI project management apprenticeship described below, formal instruction is about 8% of total program hours. The rest is mentored work on real AI projects.
What each option buys
| Broad AI upskilling | Registered AI PM apprenticeship | |
| Who it reaches | Many employees | A small number of high-potential people |
| What it builds | Everyday AI use and literacy | Ownership of AI deployments across vendors and teams |
| Format | Courses, workshops and tool training | Paid, mentored work on live AI projects, plus related instruction |
| Time | Hours to weeks | 24 months, contributing from the first week |
| Proof of outcome | Completions and usage metrics | Five stackable credentials, including CAPM and PMP, plus deployments delivered |
| Where the money comes from | The L&D budget | Mostly the apprentice’s wages, offset by state incentives |
| Retention | Varies widely | U.S. Department of Labor figures: 91% of apprentices still on the job a year later |
These aren’t competitors. Most organizations need broad literacy for many people and deployment depth for a few. The mistake is buying only the first and calling it an AI capability plan.
Funding depth without a bigger budget
An apprenticeship looks expensive if you file it under L&D. It looks different once you see where its costs actually sit.
Most of the cost is a role you’d staff anyway. The apprentice is a full-time W-2 employee, so the main expense is wages and benefits for a role a company deploying AI needs filled regardless. Program participation is scoped with the sponsor.
Registration unlocks incentives. In Maryland, the Apprenticeship Incentive Program offers up to $3,000 per newly registered adult apprentice, and the state has operated a $3,000-per-apprentice tax credit that depends on annual legislative authorization. Our guide to Maryland’s apprenticeship incentives covers the timing rules. More than twenty states run programs of their own, mapped in our guide to state apprenticeship incentives, and federal WIOA funding flows through local workforce boards in every state.
The returns are documented. The Maryland Department of Labor cites U.S. Department of Labor figures showing employers get $1.47 back in productivity for every dollar spent on apprenticeship.
The AI budget can co-fund it. If your organization’s spending looks like Deloitte’s 93/7 split, the fastest way to fund depth may sit outside L&D entirely. A small share of the technology budget, moved to the people who make that technology work, can do more for the AI program than another license tier.
A 2027 planning checklist for L&D leaders
- Split the AI line in two. Budget broad literacy and deployment depth separately, and measure them separately.
- Count your AI deployments. List the live and planned AI initiatives for 2027. Each one needs an owner, and that number sizes your depth investment.
- Find your candidates. High-potential people already on staff, AI-fluent early-career hires and transitioning veterans are all strong starting points.
- Ask the AI program owner to co-fund. Depth protects the technology investment, so part of it can come from the technology budget.
- Plan incentive paperwork with the hire. In Maryland, grant applications are due within 30 days of an apprentice’s registration, so timing matters.
- Measure depth by outcomes. Count deployments that reach production and stay there, not course completions.
Where the apprenticeship fits
On September 8, 2026, the Maryland Apprenticeship and Training Council registered the first Registered Apprenticeship in the United States dedicated to AI project management: MATC Program #2575, sponsored by Master of Project Academy and announced via Business Wire.
Over a 24-month term, apprentices complete 4,000 hours of mentored on-the-job learning on your AI initiatives and 340 hours of related instruction, 259 of them AI-specific. They earn five stackable credentials: CAPM, PMP, an AI Specialization, the Apprenticeship Certificate of Completion and 60 PDUs. Master of Project Academy handles sponsorship, instruction, certification, and registration and reporting; your organization provides the seat and a mentor. Employers in Maryland and Pennsylvania can participate today, and recognition in additional states is in progress.
If HR is weighing this against hiring or contracting, share our comparison of what each AI talent option costs. For the mechanics, see the employer playbook, or read every question about the program, answered.
Frequently asked questions
What is the difference between AI upskilling and an AI apprenticeship?
AI upskilling usually means courses and workshops that build broad AI literacy across many employees. An AI apprenticeship is a paid job with structured training: a small number of people learn to lead AI deployments by doing the work under a mentor, with formal instruction alongside, and earn recognized credentials. Upskilling raises everyday use; an apprenticeship builds the people who own deployments.
Can an L&D budget fund a registered apprenticeship?
Partly, and it often doesn’t have to carry it alone. Most of the cost is the apprentice’s wages for a role the organization needs anyway. State incentives can offset training costs, and because the apprenticeship protects the AI investment, the AI or technology budget can reasonably co-fund it.
How much do organizations spend on L&D per employee?
SHRM’s 2026 benchmarking put the median at $299 per full-time employee, down from $414 in 2025, with a median of eight training hours per employee in both years. Spending varies widely by industry and company size.
| Fund the depth, not just the literacy.
A 30-minute curriculum review scopes your vertical, your cohort size and which incentives apply in your state. |
About Master of Project Academy
Founded in 2012, Master of Project Academy has trained 500,000+ learners across 180+ countries in project management, with a 99.6% first-attempt PMP pass rate in delivered cohorts. Master of Project Academy is the registered sponsor of MATC Program #2575, the nation’s first AI Project Management apprenticeship. Learn more at masterofproject.com/p/aipmapprenticeship.
Sources
- SHRM — 2026 Learning and Development Executives Benchmarking: Developing Critical Talent
- Careerminds — Learning and Development vs. AI Spend Survey (July 2026)
- Fortune — Deloitte’s CTO on Companies Spending 93% on Tech and 7% on People (Dec 2025)
- MIT Project NANDA — The GenAI Divide: State of AI in Business 2025 (July 2025)
- Association for Talent Development — 70:20:10: Where Is the Evidence?
- Maryland Department of Labor — Maryland Apprenticeship Incentive Program
- Maryland Department of Labor — Hire an Apprentice (U.S. DOL productivity and retention figures)
- S. Department of Labor, Apprenticeship.gov — State tax credits and tuition support
- Business Wire — Maryland Registers Nation’s First AI Project Management Apprenticeship, Sponsored by Master of Project Academy (Sept 16, 2026)