Sequoia Capital is redirecting its AI strategy beyond software and coding agents, placing major bets on startups that use artificial intelligence to control and optimize physical infrastructure like energy systems, data centers, and manufacturing facilities. This emerging category, called Physical AI, represents a fundamental expansion of where venture capital sees AI's next major impact. Rather than automating digital tasks, these companies are building the intelligent control systems that will power the infrastructure underlying the AI era itself.

What Is Physical AI and Why Does It Matter?

Physical AI combines machine learning with physics-based principles to enable real-time perception, prediction, and control of complex physical systems. Unlike traditional AI software that processes digital information, Physical AI operates in the tangible world, optimizing everything from semiconductor manufacturing to data center cooling to energy grid management. DeepCtrls, a company that recently completed a Series B+ round led by CATL (the world's largest battery manufacturer) with strategic backing from Aramco Ventures, exemplifies this trend. The company develops a proprietary Physical AI Engine designed to continuously optimize and control complex systems that require constant real-time adjustments.

The practical implications are enormous. As artificial intelligence infrastructure grows exponentially, the energy demands of training and running AI models have become a critical bottleneck. Physical AI addresses this by making data centers, power systems, and manufacturing facilities dramatically more efficient. DeepCtrls already serves hundreds of enterprise customers worldwide, including major names like Tencent, ByteDance, NVIDIA, TSMC, and LG, with deployments spanning Asia, the Middle East, Europe, and North America.

"The next phase of AI will be defined by the ability to control complex systems in the real world. Energy and computing are foundational infrastructures of AI development. By connecting computing and energy through Physical AI, DeepCtrls aims to build the intelligent control layer for the infrastructure of the AI era," said Li Hui, Founder and CEO of DeepCtrls.

Li Hui, Founder and CEO of DeepCtrls

How Is Sequoia Positioning Itself in Infrastructure AI?

Sequoia Capital's recent funding activity reveals a clear infrastructure-first strategy. The firm participated in The Boring Company's $3 billion Series D funding round alongside Andreessen Horowitz and Valor Equity Partners, backing Elon Musk's tunnel-drilling venture designed to move passengers and vehicles underground while reducing surface traffic. More directly relevant to AI infrastructure, Sequoia led a $600 million Series C round for Mach Industries, a defense manufacturer founded in 2022 that develops unmanned aircraft, long-range weapons, and propulsion technology.

The firm also co-led a $550 million Series H round for Harvey, a legal AI platform, and backed Neros Technologies, an autonomous military drone manufacturer, in a $250 million Series C that valued the company at $2.5 billion. These investments signal that Sequoia sees infrastructure, defense, and energy optimization as central to AI's future, not peripheral to it.

Why Are Major Investors Suddenly Focused on Energy and Infrastructure?

The timing reflects a hard reality facing the AI industry. Training large language models and running inference at scale consumes enormous amounts of electricity. Data centers housing AI clusters require sophisticated cooling systems, power management, and energy optimization. Traditional approaches to these challenges are no longer sufficient. By applying AI to control these systems, companies can reduce energy waste, lower operational costs, and make AI infrastructure economically viable at scale.

The venture funding data underscores this shift. In the week of September 5-11, 2026, four companies each raised $1 billion or more, with significant capital flowing to semiconductor startups, defense tech, and AI infrastructure companies. Among August's 29 new unicorns, semiconductors was the second-largest sector with five new entries, and multiple companies focused on AI computing infrastructure reached billion-dollar valuations. This represents a deliberate reallocation of capital toward the physical systems that make AI possible.

How Venture Capital Is Betting on Infrastructure AI

• Energy Optimization Systems: Backing companies that use AI to reduce power consumption in data centers, semiconductor fabrication plants, and manufacturing facilities through real-time control systems and predictive analytics.
• Defense and Aerospace Technology: Investing in companies like Mach Industries and Neros Technologies that apply advanced manufacturing and autonomous systems to defense infrastructure, which requires both rapid innovation and energy efficiency.
• Semiconductor Manufacturing Tools: Supporting startups building tools and processes for semiconductor fabrication, including Fab2 and Source Foundry, which raised $500 million and $400 million respectively in recent weeks.
• AI Computing Hardware and Networking: Funding companies developing specialized hardware and networking systems for AI workloads, such as Positron and Celero Communications, which raised $500 million and $275 million respectively to build purpose-built chips and data center interconnects.

What Does This Shift Mean for AI Funding Going Forward?

The reallocation toward Physical AI and infrastructure reflects a maturing venture capital market. Early AI funding focused on software, large language models, and AI agents that automate digital work. That category remains well-funded, with companies like Cognition raising $2 billion for its autonomous coding agent Devin. However, the most sophisticated investors now recognize that AI's long-term value depends on solving the infrastructure problem. You cannot scale AI software without solving the energy, cooling, and hardware challenges that underpin it.

This explains why Sequoia and other top-tier firms are placing larger bets on physical infrastructure, defense tech, and semiconductor manufacturing. These are not tangential to AI; they are foundational. A company that can reduce data center energy consumption by even 10 percent across millions of facilities creates enormous economic value. Similarly, companies that can manufacture semiconductors faster and cheaper directly enable more AI development.

For founders and investors watching the market, the message is clear: the next wave of AI venture funding will increasingly flow to companies solving the physical world problems that AI software depends on. Sequoia's portfolio strategy, spanning energy optimization, defense infrastructure, semiconductor tools, and specialized AI hardware, demonstrates that the firm views infrastructure as the true competitive advantage in the AI era. The companies that win will not just build better models; they will build the systems that make those models economically viable at scale.