Source: S&P Global Media Portal by S&P Global.
In late 2025, a study conducted by 451 Research by S&P Global surveyed organizations’ attitudes toward cloud-native modernization and the early role of agentic AI in managing hybrid cloud workloads. At the time, the story was one of cautious curiosity: The use of autonomous AI functions, such as agentic AI, to manage cloud bursting was in discovery or proof-of-concept at most organizations; large-language-model-based code tools were beginning to reshape return-on-investment expectations; and adoption rates were broadly consistent across organization sizes.
A year later, those dynamics have diverged somewhat, with cloud-native adoption among organizations with 1,000 or more employees now at 61%, compared with 51% for those with fewer than 1,000 employees, according to a study conducted in 2026 by 451 Research. However, smaller organizations are not moving backward — they are moving more slowly, while larger organizations are accelerating. The constraints for both groups are no longer awareness, technology shortages or intent, but rather execution capacity, cost and the governance structures required to deploy autonomous AI in production environments.
The Take
AI is now a mainstream catalyst for application modernization: 70% of organizations actively modernizing say cloud-based AI tools make them more likely to migrate data to the public cloud, according to a study conducted in 2026 by 451 Research. Yet cost constraints, governance gaps and calibrated expectations about what LLM-based tools can actually address are moderating the pace. The prevailing narrative frames cloud-native modernization as a broad, advancing tide. Our data show a more stratified market, one in which larger organizations are pulling away, driven by access to capital, in-house cloud architecture expertise and the confidence to deploy agentic AI in production. Smaller organizations are not falling behind on intent. Instead, cost, legacy complexity, and skills gaps are the barriers that widen with each wave of the study. The mechanism is compounding, such that organizations already using agentic AI for cloud bursting rate its impact on workload placement speed and cost efficiency at a mean of 3.9 out of 5, further reinforcing adoption. What holds this apart is resources, not the strategy.
Value is shifting from tool availability to governance maturity — the organizations that move fastest are those that pair AI tooling with the oversight structures to run it. Funding and the availability of skills are the key conditions for this. For vendors, the implication is that the small and midmarket opportunity requires a different product motion than the organization — one that leads with cost control, managed services and prebuilt governance rather than architectural flexibility. Overall, AI tools are an accelerant for the automatable portions of modernization, and access to cloud-hosted AI is pulling migration decisions toward the public cloud in ways that were not visible two years ago.
Summary of findings
Large organizations are pulling ahead, while smaller organizations plateau. Between two studies conducted in 2025 and 2026 by 451 Research, overall cloud-native adoption held steady at the top line — approximately 52% of respondents reported cloud-native technologies in use in 2026, according a study conducted in 2026 by 451 Research. But the aggregate masks a significant structural shift by organization size. Among organizations with 1,000 or more employees, adoption stands at 61%, compared with 45% for those with fewer than 1,000 employees — a 16-point gap that was not apparent in the 2025 wave, where the study base was smaller and the difference less pronounced, according to a study conducted in 2025 by 451 Research.
Revenue sharpens the picture further. Among organizations with revenue above $1 billion, 73% report using cloud-native technologies. For those below $1 billion, the figure is 46%, while at the other end of the scale, 79% of organizations with revenue above $10 billion report adoption — the highest among all cohorts. The gap between those with and without the resources to act on a stated modernization strategy appears to be a defining feature of this market in 2026.
The availability of agentic AI is driving modernization toward the public cloud. In 2026, 30% of organizations modernizing or planning to modernize applications report using agentic AI for cloud bursting, with a further 31% in discovery or proof of concept. This contrasts with the 2025 study when agentic AI for managing cloud bursting — deploying temporary resources to handle demand spikes in hybrid or multi-cloud environments — was an early-stage phenomenon. In 2025, most respondents were in discovery or proof-of-concept, with only a minority reporting active deployment; now it is mainstream. Again, larger organizations skew toward the broader adoption pattern — 41% of those with revenue over $10 billion report using agentic AI for cloud bursting, compared with 28% for those below $1 billion. Software and IT services firms lead by industry at 37% in use, while health care organizations trail at 13%, reflecting both industry-specific risk (in)tolerance and regulatory complexity.
Where agentic AI is in use, respondents rate its impact favorably across all four dimensions tested. The mean improvement score for the speed of decision-making for workload placement is 3.94 out of 5; for the cost-efficiency of cloud bursting, 3.79; for the security posture during bursting, 3.85; and for business continuity and disaster recovery readiness, 3.80. These ratings are from current users, not aspirational assessments, indicating that more organizations are moving toward deployment. However, 61% of organizations remain in discovery, planning, or a holding pattern, suggesting that favorable user ratings have not yet translated into broad adoption in production.
LLM-based code tools are reshaping the modernization workload, but the ceiling is lower than what vendors claim. One of the most anticipated developments since a study conducted in 2025 by 451 Research has been the claimed ability of LLM-based code-generation and code-understanding tools to accelerate application modernization. The 2026 data tests that claim against organizations’ estimates. When asked what proportion of their modernization work LLM tools could address, the most common response is 41% to 60%, selected by 38% of respondents currently modernizing or planning to modernize. A further 33% estimate 21% to 40% — only 1% estimate 81% or more.
The distribution appears to matter as much as the median. Roughly 70% of respondents place LLM tool addressability at 21% to 60% of their modernization workload. This is a meaningful commercial endorsement, but it could also be seen as a ceiling signal. The vendor narrative frequently implies near-total automation of legacy code translation and refactoring; the organization estimate centers on one-quarter to one-half of the workload. Cloud-native migration work that involves mainframe-to-cloud transitions, database re-architecting and microservices extraction from monolithic applications — activities selected by 25% to 41% of modernizing organizations in the 2026 study — contains contextual, organizational and compliance complexity that LLM tools do not yet address.
Among organizations currently modernizing, 70% say cloud-based AI tools make them somewhat or significantly more likely to migrate or place data in public cloud, while 26% report no change. This suggests that AI availability is pulling data toward public cloud and accelerating cloud-native adoption. However, the effect is strongest among organizations already modernizing and does not appear to create new intent among those still only considering modernization.
Governance requirements are the final gate before agentic AI reaches production at scale. Despite favorable user ratings and growing interest in agentic AI for cloud operations, data on concerns and required safeguards reveal a consistent governance prerequisite that must be met before production deployment scales. The top concerns about using agentic AI for cloud bursting were security vulnerabilities introduced by AI decision-making (45%), followed by the complexity of integrating with existing systems (36%) and skills gaps or lack of expertise (34%). Lack of trust in AI-driven automation and compliance or regulatory risks each register 31%.
The safeguards respondents require before deploying agentic AI to manage hybrid cloud bursting form a clear hierarchy. Human-in-the-loop approvals for bursting decisions lead the list at 51%, followed by security policy enforcement at 50%, cost controls and budget guardrails at 48%, auditing and explainability of AI decisions at 47% and compliance validation tools at 45%. No single safeguard is considered sufficient; most organizations want all five. We do not believe this reflects resistance to agentic AI — instead, it is a specification of how it must be delivered. The organizations demanding these guardrails are largely the same ones that have already moved agentic AI into discovery or proof-of-concept; they are not rejecting the technology but are gating its production rollout on governance infrastructure that the market has not yet fully delivered. For vendors building agentic AI capabilities for cloud operations, the 2026 data makes the commercial priority explicit: Human-in-the-loop controls, policy enforcement and cost governance are the enabling features, not the differentiating ones.
A two-speed market offers suppliers opportunities for the next wave. The organization adoption curve for cloud-native modernization continues to steepen, and organizations currently in discovery convert to production deployment at rates consistent with historical technology adoption patterns. For large organizations, motivators from discovery to use include operational agility (47%), security improvement (41%) and customer experience (41%).
Cost and budget limitations are the primary obstacle for 39% of modernizing organizations overall, and that constraint is disproportionately greater for smaller-scale organizations. This suggests relief on cost — through managed services, cloud credits or tooling that reduces the required skills — could mitigate this impact.
The key question is whether agentic AI governance infrastructure can keep pace with deployment ambitions. Safeguard requirements — human oversight, policy enforcement and explainability — are engineering and organizational challenges that take time to resolve. If they are resolved faster at large organizations, which have the in-house security architecture and compliance functions to build or procure them, the most likely outcome is further concentration of agentic modernization benefits among those organizations that are already most advanced.
Cloud Maturity Brings Organizational IT Change
Want insights on cloud computing trends delivered to your inbox? Join the 451 Alliance.
The author(s) used an AI platform in the production of the report this blog post was based on. It was subsequently peer-reviewed, fact-checked and edited before publication. S&P Global uses generative AI to create content in accordance with our Terms of Use.
This content may be AI-assisted and is composed, reviewed, edited and approved by S&P Global in accordance with our Terms of Service.
