Orders weren’t lost, they were deferred — the structure behind summer 2026’s corporate training gap

1. Signal One — The Valve on HRD Budgets Is Tightening

The first thing to check is the money. In Gartner’s survey of about 300 CFOs, sales and IT were expected to see the biggest budget increases in 2026, while HR was squeezed the most. HR budget growth is projected to fall from 2.4% in 2025 to 0.7% in 2026, and expected growth in new hires dropped from 6% to 2%. Gartner describes this as a 'structural shift from workforce expansion to optimization through automation and AI' (as summarized by SearchlightAI).

A survey that asked Korean companies directly about training budgets picked up the tightening earlier. In August–September 2024, the Korea Productivity Center surveyed 771 workers (training managers and trainees) about next year’s training budgets. Overall, 58% expected budgets to hold and 29% expected an increase, so the total looked intact. Split by company type, though, the picture changes. 25% of large corporations expected budget cuts, versus only 8% of mid-sized and small companies. The report offered two possible reasons: large corporations had already invested heavily in training, or they were making their budgets more efficient to prepare for uncertainty. In other words, the squeeze on large corporations’ training budgets did not begin suddenly in 2026. And because large corporations account for so much of corporate training demand, the market feels far worse than '1 in 4 tightening' suggests.

Global indicators make the character of this phase even clearer. In ATD’s 2026 State of the Industry report, which analyzed data from 340 organizations, annual formal learning hours per employee rose from 13.7 to 16.7, the first increase in five years, while direct training spend per employee over the same period fell sharply from US$1,254 to US$846. For the next six months, 47% expected budgets to hold steady and 25% expected them to decrease.

In short, the budget has not disappeared. This is a phase in which learning volume is held or increased while spending is cut: companies are being asked for “more with the same money” and “by cheaper routes.” The early advance investment of the 2023–2025 AI boom is over, and H2 2026 is a period of checking that spending delivers results. The budget is not gone; the approval bar has simply been raised.

2. Signal Two — Indicators Boom, but Companies Feel Cost Pressure

The economic news in summer 2026 was bright. At the 14 July cabinet meeting, citing the semiconductor boom and policy effects, the government raised this year’s real growth forecast from 2.0% to 3.0% and unveiled the '3-4-5 Vision' (3% potential growth, a top-four exporter, and US$50,000 per-capita income) along with a second-half economic growth strategy. Semiconductors, AI data centers, and physical AI were named the three mega-projects.

But what matters is that growth is concentrated in semiconductors. In its May economic outlook, KDI noted that production costs rose rapidly as oil prices climbed because of the Middle East war, while the April consumer sentiment index fell sharply. In its H2 outlook, the Korea Capital Market Institute projected that the Bank of Korea would shift its policy stance by raising the base rate twice (50bp in total) in H2 to counter oil-driven and domestic price pressures. The H2 industry outlook from the Korea Institute for Industrial Economics and Trade (KIET) is more direct. It expects automobiles and general machinery to face continued production cost pressure from higher raw material, ocean freight, and parts procurement costs; steel to see limited profitability gains because of higher energy and logistics costs; and refining, petrochemicals, and textiles to continue facing cost and logistics burdens.

From a training manager’s perspective, this combination is the worst case. The news says growth forecasts are being revised up, yet on your own income statement costs and financing expenses are rising. And training budgets are the first line item in SG&A to be trimmed. In a rate-hike cycle, approval chains turn more conservative too. The seeming contradiction that “boom headlines and training budget tightening coexist” has a simple explanation: the industries enjoying the boom and the many companies that commission training are in different positions.

3. Signal three — AI special lectures, from boom to fatigue

Demand has clearly grown. The Korean Standards Association analyzed KSAEDU public-course enrollment data from January to November 2025 and found that demand for AI and digital transformation (DX) training rose about 290% year over year, roughly tripling. The same analysis showed that working professionals clearly preferred practical training they could apply on the job within a short period over long, theory-heavy programs.

The problem comes next. In a Chosun Edu interview (February 2026), Able Learn CEO Jina Park says many people respond that 'we took the training but can’t actually use it at work.' In particular, short seminar-style training tends to score high on satisfaction but converts at a relatively low rate into practical skills, while training designed around real work tasks shows better on-the-job conversion, she explains. The alternative she proposes is a structure of 'diagnose–design–execute–verify.'

The keynote at the HRD Special Forum, held with the Korea HRD Association, pointed to the same issue. In 2025, many large corporations ran generative AI training for employees and even produced real examples of improved work, yet the people in charge say it is “hard to feel that the whole organization has changed.” The speaker proposed a milestone: the tipping point is crossed when 50% of all employees understand the change and 15% can solve problems on their own with AI. As an example, the speaker cited a Korean company where more than 40% of all employees took part in training and 7% of them completed hands-on courses, and annual overtime fell 31%, from 32,000 hours in 2023 to 22,000 hours in 2025 (summarized by Comento).

In other words, the transition period in which demand for “just one AI talk for now” fades is exactly summer 2026. Orders have not decreased; the unit of ordering is shifting from a single talk to programs that diagnose, design, execute, and verify. This redesign takes time. Rebuilding the curriculum, segmenting participants, and agreeing on performance metrics all take time, and that time is the summer gap.

4. Signal Four — Supply Surge and Low-Cost Substitutes Push Prices Down

As demand has grown, so has supply. Corporate AI training in Korea is introduced in three ways: in-house development, outsourcing to external instructors or institutions, and government-funded courses. In-house development reflects your own data and work context as is, but content development and instructor sourcing are a heavy burden; outsourcing lets you start quickly but incurs a per-session cost; and government-funded courses carry the lowest cost burden (as summarized by Winspec). In January 2026, when it launched AI convergence training for small and mid-sized businesses, the Human Resources Development Service of Korea began supporting 90% or more of training costs — up to 95% for training offered outside the Seoul metropolitan area — across about 2,952 courses (45 training institutions) in its 'SME Talent Premium Training' (Ministry of Employment and Labor press release). This is also the result of what VOL.01 covered as 'the targeting of policy funds' showing up as real volume.

The university channel is growing too. In 2025, the AI and digital intensive courses for working professionals run by the Ministry of Education and the National Institute for Lifelong Education drew 11,683 participants across 30 universities. In 2026, the number of host institutions expands to about 38. The courses come in two formats, a short camp of about four weeks and bundled course sets. After a post-course assessment, participants receive a digital certificate (a digital badge) issued in the name of the university president. Timing is the key point: courses at newly selected universities in 2026 open for enrollment only after development finishes in August–September. Growing public supply arrives just as private orders are thin.

The result is downward pressure on prices. Standardizable areas such as AI literacy and basic usage are quickly absorbed by free and low-cost channels, pulling the market average price down. Suppliers see a deceptive boom, with volume holding while prices fall, while buyers come to expect the same training for less. That expectation soon turns into quote reviews, and those reviews in turn lead to further approval delays.

5. Signal five — not canceled, just 'pending approval' (our own observations and references)

If the signals so far describe the market as a whole, the last one comes from Nanum Consulting’s own sales opportunity ledger. It is not a representative sample of the market, but please read it as a reference showing how the previous four signals appear at the quote and contract stage in Korea (430+ records as of 4 Aug 2026).

First, new opportunities did not decline. The number of sales opportunities registered in June and July stayed at this year’s first-half monthly average. Inquiries did not dry up.

Second, contracts are being confirmed much more slowly. From August 2025 to May 2026, roughly 22–48% of contracts were confirmed within 45 days of registration each month. For opportunities registered in June, that share fell to 5%. July registrations stand at 17%, but 45 days had not yet passed when we compiled the data, so treat that figure as a lower bound.

Third, that does not mean they fell through. Among records registered in June, 27% closed as rejected or failed, no different from the 24–31% for records registered in February–May this year. Rejections are coming in at the usual rate; it is only the share that should be moving toward a win that is stuck at “in progress.” In fact, about 70% of records registered in June and July remain open. It is natural for more recent records to have more open items, but the gap is clear against the 43–56% for records registered in January–May this year.

Fourth, the nature of the replies coming back from the field has changed. Large electronics company A postponed confirming its H2 programs because “internal approval takes time,” while trading company B held provisional dates on the condition that it would “reply once internal approval is complete,” then confirmed delivery for late August. Financial IT company C scheduled its new program for November–December, and education-sector corporation D concentrated its H2 budget in August–September.

The key point is that more answers say “under approval” rather than “canceled.” Cancellation means demand has vanished; an approval in progress is only a procedural delay. When approval lines turn conservative amid rate hikes and cost pressure, even training of the same size has to go one level higher up the chain, and confirmation slips accordingly. What the data shows is not lost demand but a queue that has built up.

6. Summer Was Never Really the Off-Season (Seasonality Check, Reference)

It is easy to blame the summer gap on vacation season. Our own session records say otherwise. Converting monthly session counts over the 178 months since March 2012 into a seasonal index (overall monthly average = 100), July comes in at 114 and August at 113 — above average, if anything. The lowest months are January (54) and February (80), and the highest is November (138). In other words, over 14 years summer was not an off-season but an above-average period.

Yet July 2026 lecture volume fell to 56% of this year's first-half monthly average (as of 1 August). Seasonality does not explain a drop this steep. The vacation season may be the backdrop to the gap, but it is not the cause. If July was above average in previous years, what pulled July down this year was not the season but the five signals above.

7. Summary — The Equation Behind the Summer Lull

Combine the five signals into one formula and it looks like this. The summer gap is the product of (tighter budget valves) × (SG&A cuts in industries under cost pressure) × (orders being redesigned as one-off talks lose appeal) × (longer approvals). In this formula, seasonality is closer to a constant than a variable.

The key point is that these factors multiply. Even if each of the four has only a small effect, the combined result is large. The budget approval bar rises a notch, cost pressure triggers one more review of SG&A, planning slips a month as course design shifts from one-off lectures to an execution framework, and the approval chain gains one more level. Each adds only a few weeks of delay, but together they empty out a whole quarter.

So the corporate training gap of summer 2026 is not a disappearance of demand. Stricter budget scrutiny and a shift in the unit of procurement overlapped, pushing spending into Q3 and Q4. And deferred volume does not vanish. September through December will be the peak spending season, when that queue is released all at once. The fact that November hit 138, the year's highest, in the seasonal index we saw earlier points the same way.

8. A Second-Half Checklist for Training Managers

① Have you submitted your second-half approvals yet? September through December is the peak execution season, when deferred volume piles up. August, before the competition for instructors and schedules begins, is the best time to get approvals in.

② Have you planned it as an 'execution framework' rather than a single lecture? A multi-session structure that runs from diagnosis to training, practice, on-the-job application, and verification is built to clear the higher approval bar. (For the design approach for each course, see the Nanum Consulting programs.)

③ Have you measured success by “on-the-job application” rather than “completion rate”? — Satisfaction surveys alone cannot show whether learning is actually applied. The “measurement” covered in VOL.01 becomes the core argument for defending your H2 budget.

④ Have you changed your case to executives from 'why we need it' to 'what it will deliver'? — Approvals go faster when you cite verified change data such as shorter task times, automation rates, and report-writing efficiency.

⑤ Have you separated fundamentals from execution? — Route AI literacy and basics to government-funded, public, and university channels, and use custom programs for execution training that depends on your own work context. That split delivers budget efficiency and results together.

⑥ Have you built approval lead time into your backward schedule? — If you provisionally book instructors and venues on the assumption that confirmation will come late, you avoid losing the program after approval because no dates are left.

⑦ Have you factored the growing public-sector supply in H2 into your budget plan? New university programs in the intensive AI and digital course for working professionals open for enrollment from August–September onward.

⑧ Have you checked the risks of concentrated Q4 scheduling? When company-wide training piles into the same quarter, attendance drops. Splitting participants into groups and spreading sessions across the quarter works better for both completion and application rates.

Sources for This Issue

· SearchRight AI, “2026 Second-Half Hiring Trends”: cites a Gartner survey of about 300 CFOs (HR budget growth 2.4% to 0.7%) — View source

· Korea Productivity Center, “2025 HRD Trend Report” — next-year budget outlook item from a survey of 771 office workers (conducted Aug–Sep 2024; 25% of large corporations expect cuts) — View source

· ATD, “2026 State of the Industry”: 340 organizations (training spend per employee US$1,254 to US$846, learning hours 13.7 to 16.7) — View source

· Joint Ministries, “2026 H2 Economic Growth Strategy” (14 Jul 2026) — real growth forecast raised from 2.0% to 3.0%, three mega-projects — View source

· KDI, “Economic Outlook, First Half of 2026” (13 May 2026) — oil-driven rise in production costs, sharp drop in the April consumer sentiment index — Read the source

· Korea Capital Market Institute, “2026 H2 Macroeconomic Outlook and Key Issues” — forecast of two Bank of Korea base rate hikes (50bp total) in H2 — View source

· Korea Institute for Industrial Economics and Trade (KIET), “2026 H2 Economic and Industry Outlook” — raw material and logistics cost burdens constrain profitability in key industries — View source

· Korea Standards Association, “HRD Next: Corporate Training Trends 2026” — demand for AI and DX training up about 290% year over year (reported by Enetnews) — Read the original

· Chosun Edu, interview with Jinah Park, CEO of Ablearn (Feb 2026) — the limits of one-off talks and the “diagnose–design–execute–verify” structure — View source

· Comento, “2026 Generative AI Training Strategy”: summary of the keynote at the Korea HRD Association HRD Special Forum (tipping point and milestones) — Read the original

· Winspec Community, “How to Introduce Corporate AI Training” (June 2026) — comparing three routes: in-house development, external outsourcing, and government-funded courses — Read the source

· Ministry of Employment and Labor press release (Jan 2026) — “AI convergence training essential for small and mid-sized companies begins!” Human Resources Development Service of Korea “SME Talent Development Premium Training”: 2,952 courses, 90%+ of training costs subsidized — Read the original

· Ministry of Education and National Institute for Lifelong Education press release (Feb 2026) — 2026 plan for AI and digital intensive courses for working professionals — Read the source

· Nanum Consulting internal data — 430+ sales opportunity records (as of 4 Aug 2026) · 178 months of lecture records (since March 2012) (reference)