The Numbers That Matter
Published: July 16, 2026 | Category: Business | By Mahesh
A CEO stands in front of the board and says the company has deployed AI across most of its core processes. Revenue is up. Efficiency metrics look good on the slide. Then someone on the board asks a simple question: is the workforce actually ready for this? The honest answer, according to IBM's 2026 CEO Study of 2,000 senior leaders across 33 countries, is usually no. Only 23 percent of leaders say their workforce is genuinely ready for AI, a number that has actually fallen from 29 percent a year earlier, even though 57 percent of those same companies say they have already rolled AI out broadly.[1] Deployment moved faster than readiness. That gap is the entire story of corporate reskilling in 2026.
Almost every large employer says the same thing right now. We need to reskill our people faster than ever before. The World Economic Forum's Future of Jobs Report 2025 puts a number on that urgency. 85 percent of employers plan to prioritise reskilling their workforce. 59 percent of the global workforce will need some form of retraining by 2030.[2] Of that 59 percent, 29 percent will be upskilled in their current roles, 19 percent will be upskilled and moved into different roles, while 11 percent risk missing the reskilling they need entirely and falling behind for good.
So why, with that much urgency and that much money being spent, does readiness keep falling instead of rising?
Skills Are Expiring Faster Than Companies Can Replace Them
Corporate training used to be a stable, unglamorous function. Onboard new hires, run annual compliance modules, occasionally send someone to a leadership course. What broke that model was the sheer speed at which the underlying skills a job requires started changing. The World Economic Forum found that 39 percent of core technical skills are expected to become outdated or transformed by 2030. Even more striking, 32 percent of the skills needed for the average job in 2024 were already different from what the same job needed in 2019.[2] Five years. A third of the skill set. That is not a gradual drift. That is a fundamental shift in what "doing your job well" even means.
Fuel50's research explains why this accelerated so suddenly. AI adoption among organisations sat at roughly 50 percent for five straight years. Then generative AI arrived and pushed adoption to 65 percent in a fraction of that time.[3] The technology moved faster than any previous wave of enterprise software. Training departments, built for a slower world, simply could not keep pace. IBM's data shows the scale of what companies are now facing: 53 percent of employees will need upskilling just to perform their current role effectively between 2026 and 2028. Another 29 percent will need full reskilling into a different role entirely.[1] Combined, that is more than 80 percent of a typical workforce needing meaningful retraining within three years.
What This Is Actually Costing
$8.5 trillion. That is Skillademia's estimate of what the global skills gap costs the world economy every single year.[4] It is why reskilling has climbed to the top of the priority list for 78 percent of Fortune 500 CEOs for 2025 through 2027, ahead of cost reduction and even ahead of market expansion.
Money is following the urgency. The global corporate training market sits somewhere between $380 billion and $445 billion depending on whose methodology you trust. It is projected to climb toward $510 to $800 billion by the early 2030s.[4] Companies now spend an average of $1,420 per employee annually on training, up 18 percent from $1,220 just two years ago. Tech companies spend more than double the average, at roughly $3,270 per employee. The EdTech market serving corporate learning specifically is projected at $214 billion in 2026. Corporate learning now makes up 58 percent of all global eLearning activity, the single largest segment in the entire eLearning economy.[5]
None of that spending is small. And yet, as IBM's numbers show, readiness went down, not up.
Where the Money Goes vs Where It's Needed
Here is the uncomfortable part. Companies are not failing at reskilling because they refuse to spend money on it. Fuel50's research is blunt about this. Programmes do not fail from underinvestment. They fail because they are built around the wrong assumptions.[3] The clearest evidence of this is the mismatch between what training budgets fund and what workers actually need to develop.
Most corporate training catalogues are still weighted toward the narrow, easy-to-measure slice on the left. Technical certification is simple to design and simple to track with a completion checkbox. It is also, according to this data, solving less than a fifth of the actual problem.
The engagement numbers confirm this mismatch plays out in practice. Only 34 percent of organisations see more than half their employees actually engaging with the upskilling programmes on offer. 31 percent see fewer than one in four employees developing new skills at all. This is happening while AI-based learning platforms now power 68 percent of corporate training programmes, up from just 32 percent in 2023. Personalised learning paths have been shown to improve retention by 45 percent when implemented properly.[4] The tools exist. The content exists. The targeting is wrong.
The People Caught in the Middle
Behind every one of these statistics is someone's actual career. Anthropic's 2026 labor market study found a 14 percent drop in hiring for workers aged 22 to 25 entering AI-exposed occupations including financial analyst, project manager, legal researcher and customer service lead roles.[3] These are exactly the entry-level jobs that used to be where people learned the basics before moving up. If those doors are narrowing at the same time mid-career workers are being pushed to reskill into unfamiliar territory, the pressure is landing on both ends of the workforce at once.
D2L surveyed 996 full-time employed adults in the US and found that 75 percent expect they will need to supplement their skills to keep advancing within the next three years.[6] Half the workforce completed some form of training as part of a longer-term learning plan in 2025, up from 41 percent in 2023, so the willingness to put in the work is clearly there. But a meaningful share of those same employees still say they feel unsupported in exactly the areas that matter most for their career. Effort is rising. Support has not caught up. IBM found that 52 percent of senior leaders now say it has become harder to find employees with the right skills to execute their AI strategy. 71 percent of IT workers say they would leave for better upskilling elsewhere.[7] Nobody in the workforce is untouched by this.
The Path From Training to Actually Staying
Not everyone is stuck in this gap. A pattern shows up clearly across the companies making real progress. It starts with what happens after training ends rather than what happens during it.
Skillademia found that internal mobility programmes powered by learning and development reduce hiring costs by 30 percent. Employees who move internally into new roles stay twice as long as those hired externally into equivalent jobs.[4] Reskilling that leads somewhere specific gets used. Reskilling that sits in a learning management system with no clear next step mostly does not.
Format matters too. VR-based training grew 45 percent in 2025. Companies including Walmart, Boeing and UPS report four times faster skill acquisition for hands-on tasks compared to traditional video and slide-based content.[4] Passive content is fine for awareness. It is weak for genuine capability building, which is a big part of why completion rates and actual engagement have drifted so far apart.
And there is a broader pattern underneath both of these. Organisations with high digital maturity are 3.5 times more likely to invest seriously in reskilling than less mature peers.[7] Reskilling success is not purely a training design problem. It reflects a company's broader ability to adapt systematically, the same underlying muscle that shows up in areas like supply chain resilience, where preparing for change ahead of time is what separates companies that adapt smoothly from those that scramble.
What Actually Needs to Change
Start by auditing what skills your roles will genuinely need, not what is easiest to teach. If only 10 to 20 percent of required AI-role skills are technical, a training catalogue weighted mostly toward technical certification is solving the smaller half of the problem. Measure whether people are using new skills 60 to 90 days after training, not just whether they finished the course, since completion alone tells you people clicked through slides, not that anything changed in how they work. Attach a visible internal pathway to every reskilling programme, because an employee who knows exactly which role a new skill leads toward behaves very differently than one completing a course that goes nowhere.
And stop treating this as a project with a finish line. D2L's research frames corporate learning as shifting from a talent development initiative into core business infrastructure. Most organisations still budget it like a temporary push tied to the current AI moment. Given that a third of job skills changed in just five years, the companies that build reskilling into their permanent operating rhythm, not a campaign that wraps up in Q4, are the ones that will still be ready when the next shift arrives. And there will always be a next shift.
Common Questions
Sources
- IBM. 2026 CEO Study, 2,000 leaders across 33 countries. enterprisedna.co
- World Economic Forum. Future of Jobs Report 2025. weforum.org
- Fuel50. Upskilling and Reskilling Statistics 2026, including Anthropic 2026 Labor Market Study. fuel50.com
- Skillademia. Corporate Training Statistics 2026. skillademia.com
- Global Brands Magazine. Corporate Reskilling 2026, Fortune Business Insights data. globalbrandsmagazine.com
- D2L. Employee Training Statistics and Trends 2026, Upskill With Purpose Report. d2l.com
- WifiTalents. Upskilling and Reskilling in the IT Industry Statistics 2026. wifitalents.com