Physical AI Moves Center Stage: Why Data Center Investments Are Enabling Real-World Robotics

Mike Hogan, Chief Business Officer, GlobalFoundries 

Robots and Physical AI systems are entering a breakthrough moment, evolving from experimental demonstrations into practical tools that operate, react and make decisions in the physical world. Their rapid advancement underscores a major shift underway. AI is moving to the edge, becoming embedded in the devices and machines that surround us.

In fact, UBS predicts the market will reach up to $1.7 trillion by 2050 for humanoid robots alone. But behind this transformation lies a critical foundation: the massive investment cycle in AI data centers, which is now opening the door to Physical AI and reshaping semiconductor requirements. This is where GlobalFoundries comes in, delivering the essential semiconductor technology needed to bring Physical AI applications to life.  

The data center investment cycle 

What might look like sudden excitement around Physical AI is really the natural next step in a long‑running infrastructure build‑out. Every major leap in our industry starts this way: the foundation gets built first and then the applications explode. And that’s exactly what’s happening now, AI infrastructure is finally primed to push intelligence out of the cloud and into the physical world. 

Now, massive investments in AI data centers are driving forward a new investment cycle, but the real market opportunity is the next phase of the cycle: Physical AI, aimed at enabling billions of physical devices — from wearable devices to autonomous vehicles — to sense, think, act and communicate in real time. This cycle is accelerating at breakneck speed—market projections indicate an explosive annual growth rate of 33.49% from 2025 to 2034, an opportunity too urgent to ignore. 

The emergence of Physical AI 

This momentum signals that Physical AI is moving beyond pilots and prototypes and into scaled, real-world deployment. As data centers face mounting constraints around power and compute, Physical AI helps alleviate this pressure by shifting intelligence and decision-making to the edge directly onto devices, further accelerating this adoption. 

Physical AI continuously aggregates data from diverse sensors, processes information at varying levels of fidelity and makes decisions that are executed through motors and actuators in split seconds. These systems also operate across multiple sensor domains, enabling a true form of intelligent autonomy that allows machines to sense, think, act and communicate within the physical world. 

Real world applications taking shape 

You may not know it, but you’re likely familiar with emerging applications of physical AI today. Think of advanced driver-assistance systems, home robotics, drones, smart infrastructure, and even AI-enabled medical and diagnostic equipment. 

In markets around the world, physical AI is already being applied in areas such as mobility assistance and rehabilitation, where intelligent, on‑device systems support seniors, patients and industrial workers. These deployments show how real‑time decision‑making at the edge enables practical, scalable solutions. And as Physical AI continues to gain momentum, it is driving new and more complex demands for semiconductor innovation. 

How Physical AI workloads reshape semiconductor requirements 

As highlighted by the real-world deployments at CES this year, Physical AI workloads are fundamentally different than those driving generative AI, concentrated primarily in data centers. Physical AI introduces more complex and diverse workloads than generative AI models.  

As chip designers’ expectations rise for devices to continuously gather and process data and end users’ expectations rise for devices to execute commands and communicate in dynamic, unpredictable environments, these evolving demands are redefining semiconductor requirements. 

And here’s the breakdown of how: 

• Sense: Physical AI drives explosive growth in sensing, requiring precision analog, high‑performance RF and tightly integrated mixed‑signal solutions. GF’s FDX® platform, known for its RF performance, leakage control and high‑temperature reliability, already powers sensing in industrial automation and ADAS. As modalities expand across vision, radar, lidar and environmental sensing, GF’s precision analog, SiGe RF and integrated memory technologies provide the low‑power foundation needed to capture and preprocess real‑world data. 

• Think: Real‑time autonomy demands ultra‑low‑power, workload‑optimized compute. MIPS’ multi‑threaded RISC‑V CPUs deliver deterministic, event‑driven performance, prioritizing critical tasks and enabling instant response at the edge. Paired with GF’s energy‑efficient FDX® and FinFET technologies, these platforms push compute from centralized cloud environments into billions of devices in a distributed intelligence model. 

• Act: Executing decisions in the physical world requires high‑precision control, low latency and tight integration across motors and actuators. GF’s BCD, power GaN, mixed‑signal IP and high‑reliability CMOS support dense I/O, fast motor‑control loops and efficient actuation–core to robotics, industrial automation and emerging humanoid systems. MIPS enhances this with leading real‑time control‑loop performance for motor control, sensor fusion and decision‑making workloads. 

• Communicate: Physical AI relies on secure, low‑power connectivity across billions of devices. With leadership in RF, connectivity, SiGe and GaN for RF, GF enables everything from short‑range device links to high‑bandwidth 5G/6G and satellite communications. Combined with MIPS’ open, software‑first architecture, GF can deliver platforms that communicate quickly, efficiently and securely within distributed intelligent systems. 

Shifting requirements are driving demand for analog precision, multimodal integration, ultra-low power, optimized compute, low latency, secure connectivity, integrated memory and advanced sensing circuitry.  

Taken together, the depth of our technology and the breadth of our portfolio positions GlobalFoundries well to support the next phase of AI as it moves into the physical world. 

Delivering the technologies that power Physical AI  

As AI data center investments continue to scale, they are setting the stage for the next wave of innovation at the edge. This shift is creating a rapidly expanding opportunity for Physical AI—expected to reach at least $18 billion by 2030—as intelligent capabilities move into devices and systems across transportation, industrial automation, consumer electronics and medical technologies. Customers are now looking to bring more sensing, decision‑making, actuation and connectivity directly onto the device and the technology requirements are evolving accordingly. 

GF is building the technologies that make this transition possible. With our expanded portfolio, including ultra‑low‑power CMOS, precision analog, RF and connectivity solutions, advanced packaging and the addition of MIPS’ real‑time, multi‑threaded RISC‑V processors, we are giving customers the platforms they need to design differentiated Physical AI solutions. Our global manufacturing footprint, deep co‑design partnerships and growing pipeline of AI‑driven design wins allow us to support customers from architecture through production, helping them accelerate development and bring next‑generation intelligent devices to market with confidence. 

Innovating for tomorrow 

Today, Physical AI powers our autonomous vehicles, advanced driver-assistance systems, home robotics, drones, smart devices, even wearables and AI-enabled medical and diagnostic equipment. Looking ahead to the future, Physical AI will evolve beyond these applications, ushering in the era of humanoid robots and far more advanced autonomous systems. 

In our upcoming blog in this series, Ed Kaste, our SVP of GF’s ultra low power CMOS business, will take a deeper dive into the impact of this transition to future applications of Physical AI and GF’s role in enabling these technologies. 

Capturing the early waves of Physical AI’s inflection point 

The industry has reached a turning point: AI is no longer confined to racks of GPUs; it’s moving into the devices, machines, vehicles and robots that shape everyday life. As Physical AI proliferates, the semiconductor requirements become clearer and so does GF’s role in enabling this shift. Our next blog we’ll unpack the technical underpinnings that will define the next decade of Physical AI and how GF is positioning to lead at this inflection point.