NewsGlobalPower, Security, and Enterprise Readiness to Define AI Adoption in 2026, Say Tech CEOs at DavosPower, Security, and Enterprise Readiness to Define AI Adoption in 2026, Say Tech CEOs at Davos
Tech leaders at Davos 2026 identified power availability, cybersecurity, and enterprise readiness as the three critical factors determining AI's transition from experimentation to large-scale deployment. Energy constraints could limit AI ambitions in the US and India despite adequate chip supplies, while security concerns grow as AI systems become more embedded in business operations. Companies are moving toward production-ready agentic AI systems and neuro-symbolic approaches that combine LLM accessibility with structured enterprise knowledge.
*this image is generated using AI for illustrative purposes only.
Technology leaders at the World Economic Forum in Davos 2026 have identified power availability, cybersecurity, and enterprise readiness as the three critical factors that will determine artificial intelligence's successful transition from experimental pilots to large-scale deployment. As AI moves from promise to practice, these infrastructure and operational challenges are becoming the primary bottlenecks for widespread adoption.
AI Enters Production Phase After Years of Experimentation
Jeetu Patel, President and Chief Product Officer at Cisco, emphasized that the AI industry is entering a new maturity phase after extensive experimentation. While companies moved from simple chatbots to more advanced agentic AI systems, the focus is now shifting to production deployment.
"2026 will be the production of agentic AI," Patel stated, noting that early forms of physical AI and large world models will also begin to surface. This transition is driving investment into areas where constraints are becoming increasingly visible.
| Key Investment Areas | Primary Constraints |
|---|---|
| Infrastructure | Insufficient power, compute, and network bandwidth |
| Trust & Security | Need for reliable, safe systems at enterprise scale |
| Data Management | Machine-generated data gaps and integration challenges |
Power Constraints Emerge as Major Bottleneck
Varun Sivaram, Founder and CEO of Emerald AI, warned that energy shortages will seriously limit AI ambitions in major markets during 2026. The power bottleneck represents a critical infrastructure challenge that could determine competitive positioning in AI innovation.
"They have chips; they need power in 2026," Sivaram explained, highlighting the disconnect between available computing hardware and energy infrastructure. The scale of the challenge is significant, with plans for tens of gigawatts of data center capacity but limited grid connectivity.
| Region | Data Center Capacity Challenge |
|---|---|
| United States | 50.00 gigawatts planned, only 25.00 gigawatts can connect to grid |
| India | Similar power connection constraints |
| China | 400.00 gigawatts spare capacity expected by 2030 |
Emerald AI, backed by Nvidia, has developed power-flexible AI systems that can adjust electricity consumption in real-time. The company launched what it describes as the world's first power-flexible AI factory in Virginia, allowing data centers to connect to grids faster without increasing electricity costs for surrounding communities.
Enterprise Decision-Making Drives AI Evolution
Chakri Gottemukkala, Co-Founder and CEO of o9 Solutions, highlighted the growing complexity facing enterprises in an increasingly volatile global environment. Companies are seeking AI solutions that can support sophisticated decision-making processes beyond the capabilities of current large language models.
The next evolution involves combining the accessibility of LLMs with structured enterprise knowledge through neuro-symbolic AI. This approach aims to democratize insights by making them available to frontline teams rather than limiting access to specialists and analysts.
Security Concerns Rise with AI Integration
Jonathan Zanger, Chief Technology Officer at Check Point Software, emphasized that many AI solutions were not originally designed with security considerations. As AI systems become more deeply embedded in business operations, security gaps are creating vulnerabilities that attackers can exploit.
"In 2026, we definitely need to double down on investment to ensure AI is adopted securely," Zanger stated. Companies are responding with significantly increased budgets for AI security, with boards and CEOs making it a top priority.
| Security Focus Areas | Implementation Challenges |
|---|---|
| AI for Defense | Using AI to strengthen cyber defenses at machine scale |
| Securing AI Systems | Ensuring predictable, reliable enterprise AI behavior |
| Architecture Changes | Managing multi-data center virtual clusters |
Patel noted that while AI was initially used to strengthen cyber defenses, companies must now focus on securing AI systems themselves. Enterprise applications require predictable and reliable performance, even as AI models grow more complex and unpredictable.
Infrastructure Architecture Adapts to Scale Requirements
The evolution toward large-scale AI deployment is driving significant changes in infrastructure architecture. Power constraints are pushing the development of distributed systems where multiple data centers function as single virtual clusters.
Hyper-scalers are building clusters of hundreds of thousands of GPUs across different locations, connecting data centers where power is available to operate as unified virtual systems. This distributed approach addresses both power availability and scaling requirements as AI models continue to grow in size and complexity.
The convergence of these challenges—power, security, and enterprise readiness—represents a critical juncture for AI adoption. Success in 2026 will depend less on technological experimentation and more on solving practical infrastructure and operational challenges that enable safe, reliable, and scalable AI deployment.
18
