While Southeast Asian companies are not lacking in capital for artificial intelligence, the majority struggle to move their projects past internal IT and security reviews.
This stark reality is highlighted in the Southeast Asia findings of Adobe’s 2026 State of AI Readiness Report. The study, which surveyed 461 senior executives, reveals that 73% of regional companies are increasing their AI investments year-on-year. However, 48% remain stuck on their first AI use case or have abandoned efforts following a failed attempt.
In other words, budgets are rising while tangible results remain stagnant.
Approximately 57% of organizations are trapped in what the report terms the “Frozen Middle,” piloting or deploying AI without the coordination necessary to scale it. For agentic AI—systems capable of executing multi-step tasks independently rather than merely responding to prompts—the figure stands at 40%. Additionally, 79% of regional companies are in the early stages of AI readiness, the highest proportion across Asia Pacific.
These findings reinforce a recurring theme: Southeast Asia does not face an AI adoption problem, but rather a scaling problem.
Where the good ideas go to die
The report’s most striking statistic is the attrition rate: only 15 out of every 100 AI ideas successfully reach production.
The most significant losses do not occur in laboratories or boardrooms. Instead, about 36% of AI proposals fail to pass IT evaluation, and an additional 18% are rejected during security reviews—a rate that nearly doubles in highly regulated sectors like banking and healthcare.
While this can be framed as bureaucracy stifling innovation, it can also be viewed as gatekeepers fulfilling their duties. Many AI proposals arrive at IT departments without a clear data plan, security model, or designated owner. As previously reported by e27, unchecked agent privileges pose significant risks, and a security team that approves every agentic pilot without scrutiny would be neglecting its responsibilities.
A more pertinent question is why so many proposals arrive at IT in a state warranting rejection. The report suggests the root cause is a lack of ownership.
The ownership vacuum
Only 12% of Southeast Asian companies have an AI owner with budget authority—someone capable of both making decisions and funding them. Among the firms classified as “Leaders” by Adobe, this figure rises to 89%.
Measurement capabilities are similarly limited. Only 8% of companies comprehensively measure AI return on investment, with most tracking usage metrics like login rates rather than output quality. Furthermore, 81% of businesses report that their AI skills gap remained unchanged or worsened over the past year, aligning with trends observed in Malaysia and elsewhere in the region.
“Organisations that establish clear ownership, shared measures of success and stronger collaboration between business and IT will be best placed to move beyond isolated pilots and achieve sustainable impact,” said Jeremy Willmott, Senior Director, Digital Strategy Group Asia Pacific at Adobe.
While sensible, this is not a new recommendation. Many consultants have offered similar advice, and e27 contributors have argued that Singapore’s AI adoption challenge lies not in worker resistance but in weak execution.
What the Leaders do differently
The “Leaders” represent a small cohort: just 6.3% of Southeast Asian respondents, or 27 organizations. These firms have scaled AI across multiple workflows and established sustainable operating models. According to Adobe, they convert pilots to production at nearly four times the rate of other companies.
They also outperform the regional average by 67% on ownership and accountability, and by 59% on embedding AI into everyday processes. They maintain a “destructive failure” rate—failures causing real damage rather than lessons—below 1%. In contrast, over a third of companies repeat the same AI mistakes. Notably, 41% of Leaders cite funding employees who were already driving AI success as their single most effective initiative.
This last finding may be the report’s most practical insight. The fastest path to scaling AI often lies with the employee who quietly built a working tool in a spreadsheet, rather than through a vendor’s pitch deck. This aligns with other regional research indicating that workers are outpacing their managers in AI adoption.
Read the fine print
Before boards overhaul their AI strategies based on these figures, several caveats are worth noting.
First, the sample scope. The research, conducted by Incisiv on behalf of Adobe, focused solely on organizations with annual revenues of at least US$100 million, with 60% generating over US$1 billion. Startups and SMEs, which constitute the majority of Southeast Asia’s economy, were excluded.
Second, the Leaders’ statistics are based on a small sample of just 27 regional companies. For instance, the claim that 89% of Leaders have a budget-holding AI owner translates to roughly 24 companies. While this suggests a pattern, it is insufficient to prove it definitively.
Third, the main conclusion is somewhat circular in its methodology. The report asserts that organizational readiness, rather than technology or spending, is the strongest predictor of AI scaling. However, Leaders were initially defined by combining deployment speed with “operating model maturity.” Consequently, finding that Leaders possess mature operating models is somewhat circular. The more compelling finding is the negative one: globally, company size, industry, and reported AI spending did not predict Leader status.
Finally, Adobe sells enterprise AI tools and has a clear interest in Persuading companies that the barrier to AI value is organizational rather than technological. The company’s earlier research showed rising GenAI adoption in Asia alongside elusive ROI, making this year’s report feel like a sequel.
None of this invalidates the core diagnosis. The pattern of rising spending, stalled pilots, and lack of ownership aligns with what founders selling into Southeast Asian enterprises describe privately. Telecom providers and universities are now establishing joint initiatives, such as Singtel’s RE:AI partnership with the SIT-NVIDIA center, specifically aimed at closing the deployment gap.
For B2B AI startups, the lesson is clear: the buyer who signs a pilot is often not the person who can bring it to production. Until someone with budget authority owns the outcome, the region’s AI spending will continue to grow faster than its returns.