Introducing Labsintuition

Labs opens its Physical Intelligence lab — the frontier where intelligence leaves the screen and enters robots, machines, facilities and infrastructure.

SANTA MONICA, Calif., September 2, 2026 /Labs Newsroom/ — Labs Companies, Inc. today introduced Labsintuition, the Physical Intelligence Frontier Lab of Labs, at labsintuition.com. It is a research organization built around a single question: what happens when intelligence can act in the physical world?

Labsintelligence gives machines intelligence. Labsintuition exists to give that intelligence a world to inhabit — developing the technologies that let it sense, understand, navigate, manipulate and act across physical environments.

Most of what intelligence could do for people cannot happen through a screen. It has to happen in factories, warehouses, homes, hospitals, vehicles, farms and infrastructure — in matter, at the edge, in real time. Labs calls that frontier Physical Intelligence.

Bigger Than Robots

Robotics asks how intelligence can inhabit a body. Labsintuition asks the larger question: how can intelligence operate throughout the physical world? That includes robots, and also machinery, manufacturing systems, appliances, vehicles, wearables, devices, infrastructure and eventually whole intelligent environments.

A factory can be intelligent. So can a machine, a vehicle, a building or a network of machines — each able to perceive, reason, coordinate and respond. Robotics is one expression of a substantially larger frontier, and Labsintuition organizes that frontier by where the intelligence resides: Embodied Intelligence in an autonomous body, Machine Intelligence inside machinery, Facility Intelligence across buildings and factories, and Edge Intelligence at the device and network boundary. The ambition is not to put AI inside robots. It is to build intelligence into the physical world itself.

From Reasoning to Motion

Physical Intelligence closes a loop that software alone cannot, and Labsintuition describes it through five capabilities. Sense the world through perception across sensors, machines and environments. Understand it through physical world-models that reason about geometry, contact, causality and consequence. Navigate buildings, terrain, air, sea and land that were never mapped for machines. Manipulate matter with the dexterity to grip, lift, assemble and repair. And act, turning decisions into safe, legible, real-time motion.

Reasoning becomes decision. Decision becomes motion. Motion changes the environment, and the environment is sensed again.

1 Physical Intelligence Frontier. 3 Labs Within.

Labsintuition is organized around three specialized labs, each addressing a different way intelligence enters the world — deliberately distinct organizations designed to operate as one stack.

Labsbotics is the embodied-intelligence lab, researching autonomous robots and embodied systems that sense, understand and act in real environments. Where robotics has often started with mechanical capability and added software afterward, Labsbotics is designed around intelligence as the foundational layer. It is also the home of Labsbot, Labs' robotic agentic AI software: Labsbotics builds the embodied systems, and Labsbot gives them agency.

Labs Machines is the machine-intelligence lab. Its territory is intelligent machinery broadly — industrial and autonomous equipment, appliances, manufacturing systems and infrastructure. A machine does not need to be humanoid, mobile or conventionally robotic to be intelligent. It needs to understand its state, its environment and its objective, and increasingly to adapt and coordinate. As intelligence moves into the physical economy, some of its most consequential embodiments may look less like robots and more like the machines already surrounding us.

Factory5227 is the facility-intelligence and advanced-manufacturing lab. It builds across the hardware spectrum — machines, electronics, devices, wearables, chips, compute systems and advanced materials — and researches a more fundamental idea: that the facility itself can become intelligent. Sensors, compute, networks, robotics, energy systems and orchestration can operate as one coordinated physical system, making the building part of the intelligence architecture.

Bodies, machines and facilities — three expressions of one frontier.

Concordia Computer

At Factory5227's furthest scale sits Concordia Computer, Labs' planned multi-gigawatt unified supercomputer and a standalone supercomputer lab within Labsintuition.

Concordia begins from a different conception of computational infrastructure. The building is not housing the computer, and the campus is not housing the computer — together, they are the computer. Compute, networking, power, energy, cooling, facilities, orchestration and intelligence are being designed as parts of a single architecture that Labs calls C2I — Concordia Intelligence Infrastructure.

Compute in the Chairs, Compute in the Walls

Much of the industry already speaks about bringing compute closer to the physical world. Connected devices, instrumented facilities, sensors reporting telemetry to a computer located elsewhere — compute distributed around a building. Labsintuition means something more literal, and the difference matters for what gets built.

When Labs says the campus is the computer, it means compute in the chairs. Compute in the walls. Compute in the tables, the fixtures, the floors and the surfaces of a facility. It describes physical objects that are not endpoints reporting into a machine but participants inside one, carrying processors, memory, state and orchestration in the same architecture that runs the data halls. A wall becomes a place where computation happens. A table becomes a place where computation happens.

Labsintuition takes the doctrine further still. Elements of the physical environment are envisioned to assemble themselves and power themselves — structures, enclosures, racks, surfaces and fixtures positioned, connected, energized and reconfigured by the machines and robotics that operate a site, rather than assembled once and left unchanged for a decade. Labs Machines automates the facility, Labsbotics operates within it, Factory5227 builds the hardware, and Concordia Computer runs the compute. A facility that computes should also be a facility capable of rearranging its own computation, and that is a Physical Intelligence problem before it is a data-center problem.

What Counts as Compute

If a facility can be a computer, the question of what compute is becomes an open frontier rather than a settled definition.

Most computation resolves today to text, image, audio, video and control. Labs treats that list as incomplete. Physical Intelligence already pushes computation into perception, manipulation, navigation and force — modalities that do not reduce cleanly to tokens — and Frontier Intelligences elsewhere in Labs are exploring forms of compute whose output is sensory, spatial or experiential rather than symbolic.

One such exploration is Taste Through The Internet, an initiative being explored within YuhmmyAI, which asks whether taste and its surrounding sensory dimensions can be captured, represented, computed, transmitted across a network and reconstructed on a device — allowing a person to experience a sensory element remotely rather than only read a description of one. Sensory compute of that kind carries requirements of its own: representations that do not yet exist, models trained on non-standard modalities, latency behavior closer to interaction than to batch inference, and hardware willing to treat a sensory signal as a first-class computational object.

Labsintuition exists to give intelligence a physical world to inhabit. Sensory compute asks the inverse question — whether the physical world can be carried back through the network — and the two are the same research program approached from opposite ends.

Hardware at the Edge of the Senses

Physical Intelligence is usually described in terms of machines that move. The same frontier runs in the other direction — toward hardware worn on a person, sensing on their behalf, and reconstructing sensory experience rather than motion. If taste, touch, smell and nutrition are to become computable and transmissible, something has to sit at the boundary between the network and the body.

projectLARG — Labs AR Gadgets — is an exploration of AR-native facewear from Labs Realities. Not a display strapped to a face, but facial-computing: a wearable whose lenses sense as much as they show, running on LabsOS and carrying an XR-NXP, a Neuro-Expression Processor being explored to treat expression, attention and intent as computational inputs rather than as guesses made from a camera feed. Lens Apps describes what follows — an app and a lens becoming the same object, so what a person looks through and what they run stop being separable.

projectOPAL — Optical Palate Appetite Lenses — comes at that boundary from the sensory side, within the Yuhmmy ecosystem, and is the wearable counterpart to Taste Through The Internet. It is being explored along three directions: Taste-Technology, the instrumentation and representation of taste; Tasteimmersion, a sensory element delivered as experience rather than description; and Tasteintelligence, the reasoning that makes a sensory signal legible, personal and safe to act on.

LabsX explores the layer beneath both — fundamentally new paradigms of hardware for greater sensing across the internet, spanning taste, smell, touch and nutrition, across Labsquant, Labsneuro, Labsbio and Labsnano. The question is not how to refine an existing sensor but what a sensor should be when the thing being sensed has no established encoding at all.

These are research explorations rather than products, named here for what they imply about intelligence in physical form.

What Labs Realities Unlocks

Labs Realities creates immersive, perceptive realities across AR, VR, MR and XR — and RR, Replacement Reality — engineering the multidimensional Labsverse to power embodied interaction and future-sensory experiences. Labsintuition's interest in that work is direct: a worn device is a physical system with a body in the loop, and it obeys the same constraints as a machine that moves.

It runs continuously rather than on request. Its latency budget is perceptual, because a delay that passes unnoticed in conversation is intolerable when the result is laid over what a person is looking at. It reasons about a live scene while the scene is still live, which makes state the binding constraint rather than raw throughput. And it holds an intimate, continuous representation of a person's attention and surroundings, placing it inside the same memory and governance requirements the rest of the stack is being designed to meet.

That is why the edge and the campus are treated as one system here rather than as a device talking to a data center. Some of the work must run on the face, some near the person, and some on a campus measured in gigawatts — and deciding which, continuously and per workload, is the problem Orchestration Intelligence is being industrialized to solve.

It is also where the campus doctrine completes itself. Compute in the chairs, compute in the walls, compute in the tables — and compute on the face. Facewear is the version of that idea a person carries out of the building, the point at which an environment that computes stops being somewhere you go and becomes something you wear. Under the Computeconomy it is both at once: the most demanding consumer of compute in the system, and Native Compute already in circulation, able to contribute back into the same fabric it draws from.

Orchestration and the Edge

Intelligence in the physical world creates a coordination problem. One intelligent machine is a system; thousands of machines, robots, sensors and facilities operating at once are an orchestration problem. Labsintuition therefore includes Physical Orchestration — coordinating machines, processes, facilities and compute where action happens, as the physical-world extension of Frontier Orchestration across Labs. A robot coordinates with a machine, a machine with a production line, a line with a facility, and a facility with the energy and compute infrastructure behind it.

Physical action also cannot always wait for a distant cloud. Machines in the real world often need to perceive, decide and act within milliseconds, under changing connectivity, bandwidth and safety constraints. Edge Intelligence & Computing follows a straightforward principle: compute should exist where physical action occurs. Some intelligence will run in centralized infrastructure such as Concordia, some will run at the edge, and some will move between them.

Powering 1 Supercompany, 7 Frontier Labs, 19 Frontier Intelligences

Labs is structured as 1 Supercompany, 7 Frontier Labs and 19 Frontier Intelligences, and Labsintuition is designed as the lab where that structure meets the physical world.

Each Frontier Lab pursues a different frontier and each Frontier Intelligence a different expression of intelligence within it. Left alone, that produces 19 separate computational programs and 19 incompatible ways of describing the same primitives. A shared compute substrate is what prevents it, which is why Concordia Computer is being developed inside Labsintuition rather than beside it.

Labs describes the objective in two words. Interconnectivity — labs reach one another's models, agents, tools, memory and state across shared fabric rather than across pipelines built one pair at a time. Interoperability — what one lab builds stays legible to the next, through common orchestration semantics, state formats, runtime contracts and accounting for compute. A perception system built for one frontier should be addressable by an agent built for another without translation between them.

That is what a supercompany requires and a collection of companies does not: not merely shared services, but a shared machine.

Industrializing Orchestration Intelligence

Physical Orchestration is one instance of a larger discipline Labs intends to industrialize rather than ship as a feature: Orchestration Intelligence, the continuous coordination of many models, agents, tools, memories and states toward outcomes.

Internally, Orchestration Intelligence is how 7 Frontier Labs and 19 Frontier Intelligences operate as one system rather than as neighbors. ProjectBuzz supplies the orchestration layer, Labsintelligence supplies the intelligence, Labsintuition supplies physical execution, and Concordia provides the compute beneath all of it. Frontier Orchestration Clusters within Concordia are engineered for the shape of that work — routing, state movement, scheduling and coordination rather than raw tensor throughput alone.

Externally, Labs expects other organizations to meet the same problem without the option of building a multi-gigawatt campus to solve it. Orchestration is becoming the difficult part of applied intelligence: not obtaining a model, but conducting many of them together reliably, economically and accountably — and in the physical world, safely. Labs intends to make Orchestration Intelligence available as an industrial capability rather than hold it as an internal advantage.

Built Into a Supercompany

Labsintuition is not being built as an isolated robotics company. It inherits capability from the rest of Labs rather than recreating it — Labsintelligence supplies intelligence, ProjectBuzz supplies orchestration, Labsintuition supplies physical execution, and Labsbotics, Labs Machines and Factory5227 supply embodiment. The physical world is where the loop closes.

Labs is built as 1 Supercompany, 7 Frontier Labs because these frontiers converge. AI meets robotics, robotics meets materials, materials meet energy, devices meet infrastructure. Physical Intelligence becomes the connective layer between them, and Labsintuition sits close to that convergence point. Across the wider company, Industrial Labs spans minerals, chemicals, materials, energy, mining, construction, mobility, defense, sea, land and agriculture — distinct organizations whose frontiers increasingly intersect.

Closing the Loop

There is a further reason Labs is building the physical layer itself. Intelligence has been improving faster than the hardware that carries it, and the hardware is now the constraint. Chips are specified years before the models that will run on them exist, factories are tooled for products that change underneath them, and every generation is a bet placed on a moving target.

Labsintuition exists in part to shorten that cycle. Intelligence that designs a processor, a machine, a robot or a facility can have that design built by Factory5227, operated by Labsbotics, automated by Labs Machines and run on Concordia — and what is learned from running it feeds the next design. That is what Labs means by closing the recursive self-improvement loop in the physical layer: models improve the hardware, the improved hardware trains better models, and those models improve the hardware after that. Recursive self-improvement confined to software eventually meets a wall made of atoms. Physical Intelligence is Labs' work on the other side of that wall, and each turn of the loop is meant to be measured, reviewed and deliberately released rather than left to run on its own.

Concordia Panels, and the Open-Weights Era

Power is where the physical environment becomes intelligent first. Concordia Panels are an envisioned line of solar-powered intelligent panels — generation, storage and load management designed as computational devices rather than as electrical hardware with software attached afterward. A panel is intended to generate power, store it, understand what is drawing it, prioritize among competing loads in real time, shed and restore circuits deliberately, and negotiate with the wider facility about where energy should travel next. At scale, energy decisions and compute decisions are the same decision: scheduling a workload is scheduling power, and placing a machine is placing load.

That work sits inside a larger conviction. Labs believes open-source and open-weights artificial intelligence will define the most significant era of the field so far, and that the arrival of AGI is more likely to occur in the open than behind a single closed door. If frontier weights become broadly available, the model stops being the durable advantage, and what remains scarce is everything underneath it — compute, energy, orchestration, memory, state, fabric, and the operational capability to run the arrangement reliably for years.

Labs is also exploring Compute Exchange — CompEx — a model in which compute capacity, measured and orchestrated through C2I, could be contributed, drawn, allocated and exchanged rather than only rented from a single provider on fixed terms. Capacity is not consumed uniformly, and the AGI wave is expected to be constrained by compute long before it is constrained by ideas. Concordia Panels, Orchestration Intelligence and CompEx are each at the research or exploration stage; none are commercially available.

The Computeconomy

Physical Intelligence is one of the reasons compute demand is changing shape. Robots, autonomous systems, intelligent hardware, vehicles, machines and connected environments expand the population of things that require compute at all — a layer of physical demand that did not exist when compute was something people went to a screen to use.

That demand pushes compute outward: across devices, facilities, regions, edge systems and terrestrial data centers, and in time toward orbital infrastructure. As capacity spreads along that gradient, location becomes an economic variable rather than a deployment detail. Proximity, connectivity, energy, availability, capability and demand begin to determine where a workload runs and what a unit of capacity is worth.

Labs takes the position that computational capacity can become a new economic resource — individually owned, coordinated, exchanged and monetized wherever it exists. The Internet of Things connected the physical world. What Labs calls the Computeconomy describes that world becoming computationally and economically active, with devices, machines and facilities acting as both consumers and contributors of compute. If it holds, compute becomes as fundamental to the global economy as capital, and Exchanges join Subscriptions, Advertising and Transactions as a distinct commercial model.

Supply arrives in two forms. Native Compute is capacity already present in hardware people own — phones, laptops, vehicles, appliances, wearables, and eventually almost anything with processing capacity. Added Compute is capacity someone deliberately acquires and introduces because they want more of it. Unlike a solar panel, which cannot enter the energy economy until it is built and installed, compute can be contributed incrementally from hardware already in circulation.

The Computeconomy, Native Compute and Added Compute are research positions. No exchange has been launched, no capacity is currently offered externally, and nothing described here is available for purchase.

Stewardship and Impact

Physical Intelligence touches physical places, so Labsintuition is being built with environmental limits treated as design constraints rather than compliance work. Across its labs and infrastructure — Concordia most of all — the commitments are specific: water systems designed for closed-loop, low-withdrawal operation; electricity and power planned toward clean and firm generation, storage and grid interconnection from the first architecture review; land selected, graded and restored with the surrounding ecology as a constraint the design has to satisfy; and sea impact studied and bounded wherever a facility touches coastal or marine systems.

Frontier hardware is what makes those commitments enforceable rather than aspirational. The sensing, facility intelligence, robotics and automation Labsintuition builds instrument water, power, thermal and land use continuously across the Labs ecosystems, so consumption is reported by the infrastructure itself instead of estimated afterward. That work is carried with Labs Impact, the company-wide program that holds Labs to measurable environmental and community outcomes, and with Labsintuition Foundation, the lab's philanthropic counterpart, which directs Physical Intelligence toward applications that increase access, capability, safety and human agency. If intelligence is going to inhabit the physical world, the question is not only what machines become capable of doing, but what they cost the world to do it, and who those capabilities are built to serve.

Where This Stands

Labsintuition is a research organization. Labsintuition, Labsbotics, Labs Machines and Factory5227 have not launched the technologies described here as commercial products, and Concordia Computer remains a planned research and infrastructure program.

What Labs is introducing today is the frontier itself — its architecture, its research direction, and the organizations built to pursue it. Physical Intelligence will not arrive as a single robot, device or machine. It will emerge across bodies, machines, facilities, networks, compute infrastructure and environments. The ambition connecting all of it is larger than robotics: build intelligence capable of entering the physical world, and build a physical world capable of becoming intelligent.

Labsintuition, LLC is lab⁷ of Labs Companies, Inc. — 1 Supercompany, 7 Frontier Labs. Learn more at labsintuition.com.

Media: press@labscompanies.com

Introducing LabsintuitionBuilding Physical Intelligence Across Many Frontiers