Industry executives and experts share their predictions for 2026. Read them in this 18th annual VMblog.com series exclusive.
By VJ Sahi, Partner and Senior Vice President at Clark Street Associates
The past two years have rewarded those with an ambitious AI implementation strategy. In 2026, those with a disciplined AI execution will be rewarded.
There has been a historic surge in investment in AI infrastructure, and that initial hype is settling down. Now, we’re entering a new era defined by whether the massive capital deployed can produce durable, defensible returns.
In 2024 and 2025, an “AI strategy” was new and held a different meaning. Companies decided to expand simply for the sake of expansion, and they rushed to be a player on the field. That meant more models, more pilots, and more infrastructure. While this approach helped them learn quickly, it also created a gap between promise and payoff, because budgets began to tighten and boards demanded proof of noticeable ROI. With ROI still top of mind, 2026 will be marked by a shift from experimentation to execution.
The AI boom is not ending; it’s maturing. And with maturity comes scrutiny.
Moving from Expansion to Outcomes
In the next phase of AI adoption, the definition of success will be put to the test. Previously, companies were considered successful if they were using AI. Period. Now, with almost every company deploying AI throughout its technological ecosystem, it is understood that the money invested in AI must yield a return. This means companies will start to rationalize their AI portfolios, consolidate vendors, and retire redundant models.
This recalibration is inevitable. While it’s a fact that these AI systems are incredibly powerful, power alone won’t bring success. Organizations will increasingly ask whether AI improves margins, reduces operational friction, or unlocks revenue in scalable ways. It will be up for debate whether the models that cannot answer these questions remain.
For investors, this shift is equally important. Capital will begin to shift away from speculative infrastructure buildouts and focus more on companies that can demonstrate how AI can be successfully integrated into real-world workflows. They’re looking for the models that won’t be the biggest, but instead the ones that produce the clearest returns.
Additionally, this change will divide the early adopters who invested in strong data foundations and those who rushed ahead without them. AI systems require a solid foundation of clean, ready-to-use data to achieve high-quality outcomes. Poorly governed data will quickly hit performance ceilings. Those organizations that invest in governance and trust will find that using their capital to scale AI is a worthwhile investment.
The Rise of Sovereign AI
While enterprises demand returns, governments are rethinking their dependence.
Another major development shaping AI in 2026 will be the acceleration of sovereign AI initiatives. According to the Organisation for Economic Co-operation and Development (OECD), more than 70 countries have national AI strategies, with a growing emphasis on domestic computing, localized data control, and public sector models.
Having an AI strategy is not just about regulation, but also about harnessing power. Governments across the globe are emphasizing how AI dependence can be a strategic risk, as it limits their ability to enforce policy and remain competitive with foreign adversaries. As a result, we can expect to see the continued strengthening of AI initiatives inside countries, particularly in the United States, as they compete with rivals like China.
Infrastructure, Not Illusions
The entire AI landscape is being reconfigured in 2026, with a broader focus on outcomes and sovereignty. And with these focal shifts, attention is also drawn toward infrastructure choices.
Companies are deciding what AI models will give them a strategic advantage, and in response, will optimize for reliability, governance, and cost discipline rather than raw capacity. Unlike traditional enterprise applications, AI workloads are far more compute-intensive and power-dense, making them a primary driver of data center redesign. This will prompt enterprises to favor hybrid and on-premises environments, enabling them to manage costs, latency, and compliance more effectively.
Infrastructure is no longer a technical decision but will become a business one. AI environments that can’t be governed, audited, or aligned with regulatory requirements will be seen as a liability rather than an asset.
A Guided AI Future
With expectations for AI changing by the minute, the defining characteristic of AI in 2026 will be accountability. Organizations are entering a new era where the outcomes of their AI models will be evaluated on their legitimacy and having a result that is either incorrect or of low quality will put them at a disadvantage. Not only that, but it could put them in a legally difficult situation. This is why we will see companies prioritize governance and trust over model sophistication and size.
In 2026, AI will continue to evolve at a rapid pace, but this time with a clear purpose. Enterprises will make this shift from treating AI as a marketing signal to a core operational capability, and the governments that invest early in building their domestic AI infrastructure will recognize that technological independence is a form of economic power.
The era of asking what AI might do is ending. The next question is even more important: what is it delivering?
##
ABOUT THE AUTHOR
VJ Sahi is Partner and Senior Vice President at Clark Street Associates. He is primarily responsible for Clark Street’s government relations within DOD, Department of Commerce and on Capitol Hill, assisting commercial companies in areas of advanced electronics, semiconductors, quantum computing and machine learning.
