Intel and Outsight Announce Strategic Collaboration to Bring Physical AI–Powered Spatial Intelligence to the Enterprise Edge

Frequently Asked Questions

• Can LiDAR-based spatial AI run on a CPU without a dedicated GPU at the edge?

  Yes. Intel's Xeon 6 SoC includes Advanced Matrix Extensions (AMX), a hardware acceleration block built into the CPU specifically for matrix-heavy AI inference workloads. This allows 3D point cloud processing and entity tracking to run at sub-50ms latency without a discrete GPU. Outsight's SHIFT platform demonstrated exactly this capability when integrated into Google Distributed Cloud Edge powered by Intel Xeon 6 SoC, showing that infrastructure-based Physical AI can operate at the enterprise edge without dedicated GPU hardware. The practical consequence for operators is lower rack power draw, fewer hardware SKUs to qualify and maintain, and a total cost of ownership profile that supports deployment across many distributed sites rather than a small number of GPU-equipped data centers.

• What is Google Distributed Cloud Edge and why does it matter for physical AI deployments?

  Google Distributed Cloud Edge is a managed hardware-and-software stack that extends Google Cloud's control plane to on-premises or near-premises locations. For physical AI workloads, this means the inference pipeline runs at the site (sub-100ms latency, no cloud round trips), while configuration, software updates, and security policies are managed centrally through Google Cloud. Enterprise operators running dozens of airports or factories benefit because they avoid building separate edge management infrastructure for each site. Outsight's SHIFT platform deploys as a managed workload on Google Distributed Cloud Edge via the Intel Xeon 6 SoC integration, allowing real-time 3D perception from infrastructure-mounted LiDAR sensors to run directly at each physical site without requiring a dedicated cloud uplink for every processing decision.

• How does infrastructure-mounted LiDAR give autonomous mobile robots a wider view than their onboard sensors?

  A robot's onboard sensors see only what is within line of sight and within the sensor's detection range, typically a narrow forward cone. Infrastructure-mounted LiDAR covers the entire floor plan continuously, tracking every person, vehicle, and object with a persistent anonymous ID. Outsight's approach places LiDAR in the infrastructure rather than on the robot itself, feeding a real-time Motional Digital Twin that the robot can query to know what is happening around a blind corner, behind a shelf, or across the warehouse before it physically reaches that position. This shared spatial map extends effective perception radius from a few meters to the full premises, removing the fundamental blind-spot limitation that onboard sensors cannot solve on their own.

• What is multi-modal data fusion in a spatial intelligence platform?

  Multi-modal fusion means combining position and trajectory data from LiDAR with attributes from other sources: camera streams, access-control logs, business databases such as point-of-sale systems, flight information (AODB), or Wi-Fi dwell signals. LiDAR provides the spatial backbone and the persistent anonymous ID for each tracked entity, while the other sources attach contextual attributes to that ID. The SHIFT platform by Outsight operationalizes this approach through open integrations, allowing operators to enrich a real-time 3D scene with data layers from existing site infrastructure. The result is a richer situational picture than any single sensor type can produce alone, without requiring any one sensor to do everything.

• How is 'anonymous by definition' different from video anonymization software?

  Video anonymization processes images or video frames after capture and blurs, pixelates, or masks faces before storage. The raw identifiable image still exists in memory or on the network during that processing step, creating a window of exposure. LiDAR-based capture never records pixels at all: the sensor emits laser pulses and records return distances, producing a 3D point cloud that contains shape and motion but no facial geometry or license plate text. There is no post-processing anonymization step because no biometric data was captured in the first place. Outsight describes this property as "anonymous by definition" and builds it into the SHIFT platform.

• What does 'continuous individual tracking' mean in a large venue or factory, and why is uninterrupted ID important?

  Continuous individual tracking means each person or object is assigned a single anonymous ID the moment it enters a monitored area, and that ID persists without interruption until the entity exits, even when moving between overlapping sensor zones, passing through crowds, or briefly leaving one sensor's field of view. Outsight's SHIFT platform handles this through infrastructure-based 3D LiDAR perception, where the Motional Digital Twin maintains consistent object identity across the entire sensor network in real time. Uninterrupted ID matters because analytics that depend on journey time, dwell, or behavioral sequence (such as detecting a worker who enters a restricted zone, leaves, and re-enters) require a consistent identity reference throughout. Fragmented IDs from handoff gaps corrupt those calculations and generate false alerts.

• When and where did Intel demonstrate the physical AI edge solution at Google Cloud Next 2026?

  Intel and Outsight announced their strategic collaboration on April 21, 2026, from Las Vegas and Paris. A live, full-scale demonstration of the integrated physical AI solution was showcased at the Intel booth (#6901) during Google Cloud Next 2026, held April 22 to 24 at the Mandalay Bay Convention Center in Las Vegas. The demo ran Outsight's Shift platform on Intel-powered edge infrastructure within the Google Distributed Cloud framework, showing real-time detection, classification and tracking of people and objects.

• What role do Intel AMX Advanced Matrix Extensions play in running the Shift platform on Xeon 6 SoC?

  In the Intel and Outsight collaboration, Outsight's Shift platform runs on Intel Xeon 6 SoC processors that include AMX (Advanced Matrix Extensions), which accelerate CPU-based inferencing. Combined with Outsight's 3D real-time processing platform, this delivers turnkey physical AI capabilities directly at the edge of enterprise networks. Running natively on these processors enables cost-effective, scalable 3D AI inference, which lowers total cost of ownership and makes deployment across thousands of locations practical.

• What did the Intel live demo at Google Cloud Next 2026 show about crowd density, queues, and Motional Digital Twins?

  The Google Cloud Next 2026 demonstration showed how Outsight's platform processes millions of 3D data points per second to detect, classify and track individuals and objects in real time, while preserving privacy through anonymous, non-image-based sensing. Key capabilities on display included real-time crowd density monitoring, queue management analytics, flow visualization through a live Motional Digital Twin and predictive operational insights. Everything ran on Intel-powered edge infrastructure within the Google Distributed Cloud framework.

• What are Motional Digital Twins in the Intel physical AI collaboration?

  In the Intel and Outsight solution, Motional Digital Twins are real-time 3D digital replicas of physical flows that serve as the operational interface for end users. Through them, operators of airports, stations, factories and venues monitor, analyze, predict and optimize their environments. The Motional Digital Twin continuously tracks every person, vehicle and object across the premises and assigns each a unique anonymous ID maintained over time, so the whole site becomes a shared, real-time representation of reality available to operators and robots.

• How does situational awareness link event history and external data to tracked robots and people?

  Situational Awareness is one of the three foundational capabilities enabled by the Intel and Outsight solution. It means perceiving the environment globally, beyond the line of sight of each individual robot, and associating external data, event history and attributes with each tracked entity. Because every person, vehicle and object keeps a unique anonymous ID over time, this context stays attached as they move across the premises. Operators and mobile robots can then understand what is happening site-wide in real time.

• How does the Intel-powered LiDAR solution predict future operational states across a factory or airport?

  The third foundational capability of the Intel and Outsight solution is Actionable Insights and Prediction. Building on continuous individual tracking and site-wide situational awareness, the platform interprets situations at scale, derives decision-ready intelligence and anticipates future states across the entire premises. For operators, predictive operational insights were among the capabilities shown in the live demonstration at Google Cloud Next 2026. Mobile robots gain premises-wide context that enables safer navigation, smarter path planning and coordinated fleet operations.

• Why does deploying the Shift platform through Google Distributed Cloud Edge simplify multi-site rollouts?

  Integration into Google Distributed Cloud Edge lets enterprises deploy Outsight's Shift platform as a managed, cloud-connected edge workload. Customers benefit from streamlined operations, centralized management, automated updates and enterprise-grade security within Google Cloud's security and scalability framework. Combined with CPU-based inference on Intel Xeon 6 SoC, which lowers total cost of ownership, this model is designed for deployment across thousands of locations, from airports and stations to factories and smart cities.

• How did Raul Bravo describe the significance of the strategic collaboration with Intel?

  Raul Bravo, President Founder of Outsight, called the collaboration with Intel a major milestone in making physical AI accessible at scale across critical infrastructure worldwide. He explained that integrating the Shift platform into Google Distributed Cloud Edge, powered by Intel processors, lets operators of airports, stations, factories and smart cities deploy a real-time 3D replica of everything moving across their premises. It also provides global situation awareness to mobile robots as a standard component of their edge infrastructure.

• What did Intel's Bhupesh Agrawal say about CPU-based 3D AI inference in this partnership?

  Bhupesh Agrawal, General Manager, Private 5G and Enterprise AI at Intel, said enterprise AI is about bringing intelligence closer to where decisions are made, at the edge. He stated that the collaboration with Outsight shows how Intel's edge computing platform, combined with best-in-class physical AI software and the Google Distributed Cloud, enables organizations to transform physical operations with real-time, CPU-based 3D AI inference. He described it as the scalable, cost-effective solution enterprises need to unlock the full potential of physical AI.

• How widely is the Spatial Intelligence platform deployed, and what recognition has it received from Gartner and award bodies?

  Outsight's Spatial Intelligence solutions are deployed at scale in major airports, train stations, stadiums, industrial sites and smart cities across five continents, and more than 280 million annual passengers travel under the real-time measurement of Outsight's Spatial Intelligence Platform. The company has received seven citations as an emerging leader from Gartner in the digital twin and spatial computing markets. Its awards include the CES Best of Innovation Award, the Frost and Sullivan Global Technology Innovation Leadership Award and the Edge AI and Vision Product of the Year.

• Does real-time people tracking require a GPU or can it run at the edge?

  Real-time 3D tracking can run efficiently at the edge on standard processors when the software is optimised for it. Outsight's Shift platform runs on Intel CPUs at the edge and has been integrated into Google Distributed Cloud Edge powered by Intel Xeon 6 SoC, as shown at Google Cloud Next 2026. Processing close to the sensors keeps latency low and sends only compact results to the cloud. This makes large deployments more practical and energy-efficient.