'We have maybe 20 months' to rebuild for AI agents, Meta's infrastructure VP tells VB Transform 2026

Meta's Vice President of Infrastructure has issued a stark timeline for the industry, stating at the VB Transform 2026 event that there are perhaps 20 months to rebuild infrastructure to accommodate AI agents. This warning highlights a critical shift toward agentic AI, which is expected to fundamentally change how technology systems are structured. This development coincides with reports from KPMG suggesting that agentic AI will significantly reshape network demand, creating a new set of requirements for how data and connectivity are managed across global digital landscapes. Multiple industry outlets are tracking these systemic pressures. VentureBeat focuses on the specific time constraints mentioned by Meta's leadership during the Transform 2026 conference.

Meanwhile, IDC has released survey results flagging the emergence of networking bottlenecks that could hinder the deployment of agentic AI. RCR Wireless News emphasizes the perspective of KPMG regarding the reshaping of network demand, while China Daily notes that the creation of AI requires more than just the AI software itself, pointing toward the physical and systemic foundations necessary for these tools to function. Contextualizing the crisis, the coverage indicates that the primary constraints are shifting away from raw processing power. According to EE Times India, the next major bottleneck facing the industry is power rather than compute. This suggests that while the ability to process data exists, the energy required to sustain agentic AI systems is becoming the dominant hurdle.

The collective reporting underscores a transition from focusing on model development to focusing on the physical infrastructure, power grids, and networking hardware that must support autonomous agents acting on behalf of users. Future developments will likely center on whether infrastructure can be scaled within the 20-month window cited by Meta. Monitoring will be required to see if the power bottlenecks identified by EE Times India are mitigated or if they exacerbate the networking issues flagged by the IDC survey. Additionally, the industry will be watching for how the reshaped network demand predicted by KPMG manifests in actual hardware upgrades. The focus remains on the physical layer of technology, as the shift to agentic AI necessitates a total rebuild of the supporting environment.