The AI Spending Paradox — Spending More, But Still Unable to Prove ROI
2026 is becoming the "year of truth" for Enterprise AI worldwide.
The latest findings from McKinsey’s State of AI make it clear that 88% of organizations are already using AI in some form, and 71% plan to increase AI budgets this year. But the striking figure is that only 39% report significant bottom-line impact from their AI investments.
That means nearly two-thirds of organizations investing in AI are still pouring money into it without being able to prove that it is worth the cost.
| Indicator |
Figure |
Source |
| Organizations already using AI |
88% |
McKinsey State of AI 2025 |
| Planning to increase AI budgets in 2026 |
71% |
McKinsey State of AI 2025 |
| Seeing real bottom-line impact |
39% |
McKinsey State of AI 2025 |
| Global AI spending in 2026 |
$2.5 Trillion |
Gartner |
| Organizations still in the pilot stage |
54% |
Deloitte State of AI 2026 |
This contradiction is exactly why 2026 is becoming the defining moment between organizations that will move ahead and those that will be left behind.
$2.5 Trillion — The Number That Should Make Everyone Stop and Think
Gartner forecasts that global AI spending will reach $2.5 trillion in 2026, up more than 30% from 2025. This massive investment is continuing to flow into Generative AI, Agentic AI, AI Infrastructure, and AI-powered SaaS.
But the question every executive must answer is: What is every dollar invested actually returning?
If 88% are using AI but only 39% are seeing real results, that implies hundreds of billions of dollars are being spent without a clear path to ROI.
For Thai organizations, this represents both a risk and an opportunity. The risk is following the trend and investing without a measurable outcomes plan. The opportunity is learning from the mistakes of the global market and skipping unnecessary steps.
AI Leader vs AI Laggard — A Gap That Widens Every Day
McKinsey clearly separates organizations into two groups: AI Leaders, which already generate measurable business value from AI, and AI Laggards, which remain stuck in the experimentation phase.
The difference between these two groups is not budget or technology. It lies in how they think and how they operate.
| Area |
AI Leader |
AI Laggard |
| Starting point |
Begins with a clear business problem |
Begins with technology they want to try |
| Measurement |
Sets KPIs and baselines from day one |
Measures later, or not at all |
| Team structure |
Cross-functional: business + technical + data |
Siloed: IT handles AI, business waits for results |
| Scope |
Scales from pilot to production within 6 months |
Stuck in pilot for more than 12 months |
| C-suite support |
CEO/CFO directly involved |
Left solely to CTO/IT Director |
| Revenue impact |
Average revenue increase of 6–10% from AI |
No visible revenue impact yet |
McKinsey data shows that AI Leaders deliver 2.5x higher revenue growth than AI Laggards, and this gap is widening every quarter because AI has a compounding effect — the more it is used, the more data is generated, and the faster it improves.
The 5 Most Common Failure Patterns
Analysis across multiple sources shows that organizations investing in AI without seeing results often fall into these patterns:
1. Pilot Purgatory — Endless Experimentation
Organizations launch multiple AI pilots at once, but none are ever scaled into true production. Everything remains a "test project" with no end in sight. Deloitte found that 54% of organizations are still in this state.
2. No Clear Success Metrics
They start AI projects without defining what success looks like from the outset. There is no baseline data before launch, so once the project is underway, no one can clearly say whether things have improved.
3. Siloed AI Initiatives
Each department builds its own AI efforts independently, with no connection, no shared data, and no shared learnings. The result is duplication, wasted budget, and no cross-functional synergy.
4. Choosing the Wrong Use Case
They invest in AI for low-impact tasks or select problems that are too difficult for the organization’s current AI maturity. As a result, projects take longer and deliver less value.
5. Lack of Change Management
They focus only on technology and forget that people across the organization must change how they work. If people do not use the system, even the best technology will not generate ROI.
What AI Leaders Do Differently
Organizations that succeed with AI do not necessarily have better technology. What they clearly have is a different mindset and process:
They Start with the Business Problem, Not the Technology
AI Leaders never begin with the question, "What can we do with AI?" They begin with, "What business problem hurts the most, and where can AI realistically help?" Starting with a real problem gives everything that follows a clear direction.
They Measure from Day One
Before they begin, these organizations will:
- Define baseline metrics to establish the current state
- Set clear, measurable target metrics, such as reducing processing time by 40% or increasing conversion rate by 15%
- Define a timeline for evaluation — 30 days, 90 days, 180 days
- Agree on go/no-go criteria to determine when to stop, pivot, or scale if results do not meet expectations
They Build Cross-Functional Teams from the Start
Successful AI projects are never delivered by IT alone. The working team includes:
- A Business Owner who understands the problem and context
- A Data Team that prepares and maintains data quality
- A Technical Team that designs and develops the system
- End Users who will actually use it and provide feedback early
They Scale Fast, Fail Fast, Learn Fast
Instead of running a 12-month pilot before making a decision, these organizations work in short cycles — 4–8 week pilots followed by measurement. If the project meets the criteria, they scale immediately. If not, they pivot or stop. They do not waste time on projects with no future.
Where Thai Organizations Stand in This Picture
For Thai organizations, the current situation contains both concerns and opportunities.
Challenges Specific to the Thai Market
- A more severe talent gap than the global average. Thailand still faces a major shortage of AI talent, especially at the AI Engineer and AI Product Manager levels.
- Low data readiness. Many organizations still have fragmented data spread across Excel files, legacy systems, or paper-based processes.
- Evolving regulations. Stricter PDPA enforcement and the upcoming AI Act mean organizations must consider compliance alongside AI adoption.
- Budget constraints — mid-sized organizations in Thailand often have more limited AI budgets than regional competitors, which makes use-case selection even more critical.
Opportunities Still Wide Open
- Late-mover advantage — organizations can learn from the mistakes of earlier markets and skip unnecessary steps.
- Government support — Thailand is positioning itself as an ASEAN AI Hub, backed by 5GW of Data Center capacity under development.
- Lower AI costs — API fees, compute costs, and SaaS-based AI tools continue to decline, making ROI easier to achieve.
- Use cases with clear impact in the Thai context — Customer Service Automation, Thai-language Document Processing, and Demand Forecasting for retail and manufacturing.
A Framework for Evaluating AI ROI — Look Across All 4 Dimensions
Accurately evaluating AI ROI requires looking across all four dimensions, not just "how much was invested versus how much came back."
Dimension 1: Financial Impact
- Revenue Uplift — Does AI directly increase revenue, such as through higher conversion rates or more accurate upselling?
- Cost Reduction — What costs does it reduce, such as manual work time, error rates, or processing expenses?
- Cost Avoidance — What future costs can be avoided, such as not needing to hire additional staff as workload grows?
Dimension 2: Operational Efficiency
- Throughput — How many items can be processed per day compared with before AI?
- Cycle Time — How long does the full process take from start to finish?
- Quality — By how much has the error rate decreased, and by how much has accuracy improved?
Dimension 3: Strategic Value
- Competitive Advantage — Does AI enable the organization to do something competitors cannot?
- Speed to Market — How much faster can new products or services be launched?
- Data Asset — Does the data generated through AI become a strategic asset for the organization?
Dimension 4: Risk & Compliance
- Cost of Inaction — If the organization does not act on AI today, what will the cost be in three years?
- Regulatory Readiness — Is the AI being used aligned with PDPA and upcoming regulations?
- Operational Risk — Does AI reduce human error, or does it create new risks that must be managed?
The key question executives should ask before approving any AI project is:
"How will we know this project has succeeded — and how many days will it take us to know?"
If that question cannot be answered, the organization is not ready to start.
2026 Will Decide Everything
The gap between AI Leaders and AI Laggards is widening at an accelerating pace because AI has a compounding effect — organizations that begin seeing results gain more data, improve models faster, and build competitive moats that become harder to catch every day.
What Thai organizations should do now is not rush to buy expensive AI tools, but instead:
- Review existing AI initiatives — Which projects have real measurable impact, and which are still endless pilots?
- Define clear success metrics — Every AI project should have measurable KPIs from day one.
- Build cross-functional teams — Stop leaving AI entirely to IT.
- Evaluate AI ROI across all 4 dimensions — Financial, Operational, Strategic, and Risk.
- Choose use cases that are both high-impact and feasible — Better than running 10 small projects that no one remembers.
2026 is the year the truth will be revealed — organizations that are prepared will move ahead, while those that hesitate will be left behind.
Enersys helps Thai organizations build AI strategies with measurable outcomes — from assessing AI readiness, selecting the highest-value use cases, and defining a metrics framework to scaling from pilot to production.
Sources
Consult AI Strategy with Enersys →
