DSTL - Audio in. Alpha out.

dstl

TERMINAL

LIBRARY

Gecko Robotics CEO, Jake Loosararian on AI and Physical Robots

Join

Gecko Robotics CEO, Jake Loosararian on AI and Physical Robots

Jan 23, 2026

All-In Podcast

16:17

94K Views

THESIS

Specialized robotics serve as the essential data ingestion layer for the physical world, enabling AI foundation models to deliver high-ROI operational improvements in critical defense and energy infrastructure.

ASSET CLASS

SECULAR

CONVICTION

HIGH

TIME HORIZON

3 to 5 Years

PREMISE

The Physical World Data Deficit

The current AI revolution faces a structural bottleneck: while LLMs effectively index the internet, there is no comparable data corpus for the 'built world.' Critical industrial sectors like defense, energy, and manufacturing lack the digitized, high-fidelity datasets required to train models on asset health, welding quality, and structural integrity.

MECHANISM

Robotics as the Industrial Nervous System

Gecko Robotics deploys purpose-built, sensor-laden robots to physically traverse and inspect infrastructure. This acts as a 'nervous system' that captures proprietary, granular data from the physical environment, creating the necessary training data to build foundation models specifically for industrial maintenance, manufacturing, and decision-making.

OUTCOME

Tangible Industrial ROI and Speed

By bridging the gap between digital AI and physical reality, this approach unlocks massive efficiency gains—such as a 90% increase in manufacturing speed for defense assets and the ability to extend the useful life of energy infrastructure—generating immediate commercial value where generalist models fail.

NECESSARY CONDITION

Continued access to critical physical infrastructure for data collection.

There's going to be need to be an application layer and operational excellence in the field... that's exactly what we are with a nervous system to pull all this information from robots together and then you can build and use AI models on top of that to use the information data to then also begin to take actions back into the real world.

85:34

RISK

Steel Man Counter-Thesis

While Gecko Robotics effectively captures a unique 'Physical World' dataset, its reliance on bespoke hardware creates a linear scaling drag in an exponential AI market. Even with a data moat, the rapid commoditization of general-purpose humanoid hardware and multi-modal learning capabilities could allow larger capitalized competitors (Tesla, Figure) to flood the zone with 'good enough' inspection units, relegating Gecko to a niche high-end service provider rather than the dominant 'nervous system' for the built world.

RISK 01

Hardware Manufacturing Bottleneck

THESIS

The CEO explicitly identifies the physical manufacturing of robots as 'the big issue'. Unlike software-only AI companies that scale infinitely with compute, Gecko's data advantage is constrained by its ability to physically build, deploy, and maintain complex sensor-laden hardware units. A failure to scale manufacturing efficiently stalls the data ingestion engine.

DEFENSE

The CEO acknowledges this is the current primary focus for the company's operational excellence over the next 5 years, though specific solutions (e.g., outsourcing vs. internal vertical integration) were not detailed.

RISK 02

Generalist Model Leapfrog (The 'Optimus' Threat)

THESIS

The risk that general-purpose humanoid robots (e.g., Tesla Optimus), utilizing vision models and public training data (web videos, manuals), will master trade skills like bricklaying or welding without needing Gecko's specialized sensors. If foundation models generalize faster than anticipated (CEO estimates ~3 years ), specialized data moats could depreciate.

DEFENSE

The CEO argues that the specific, high-fidelity data required to assess asset health (e.g., internal corrosion, microscopic weld integrity) does not exist on the internet or in video format. Physical access via specialized sensors is required to create the 'ground truth' dataset.

RISK 03

The 'Data Tax' on Customer ROI

THESIS

The CEO admits that industrial customers (Energy, Defense) will not pay simply to train Gecko's AI models (e.g., 'strapping GoPros on arms'). The business model requires that the immediate service provided (inspection) yields positive ROI independently. If the cost of robotic deployment exceeds the immediate operational savings, the data pipeline—which is a byproduct of the service—evaporates.

DEFENSE

The CEO claims the ROI is 'easy to underwrite' because they focus on massive GDP-driver problems like extending asset life and increasing manufacturing speed by up to 90%, effectively subsidizing the data collection.

ASYMMETRIC SKEW

Asymmetric Upside: The downside is protected by a viable industrial service cash flow, while the upside offers 'Platform' potential if they successfully monopolize the physical health dataset of critical infrastructure.

ALPHA

NOISE

The Consensus

The market is fixated on 'Internet AI' (SaaS, digital-only models) and Generalist Humanoid Robots (e.g., Tesla Optimus) focused on visible, low-ROI domestic tasks like laundry, assuming training data can be scraped from public video repositories.

Train robots using vast public corpora of internet videos (e.g., YouTube tutorials on bricklaying) to achieve generalizable physical skills through visual imitation.

SIGNAL

The Variant

The true structural opportunity is in the 'Built World' (Defense, Energy), where critical data is offline and invisible to the internet. The goal is not just labor substitution, but creating a 'nervous system' for industrial assets to capture a proprietary data monopoly on structural health and manufacturing quality.

Visual data is insufficient for high-stakes industry (e.g., submarine hull integrity). You must deploy sensor-laden robots to physically contact and inspect assets to generate the 'Ground Truth' dataset that doesn't exist online. This proprietary data is the prerequisite for training valid physical AI models.

SOURCE OF THE EDGE

Privileged Data Access (Proprietary 'Built World' Sensor Data) & Geographic Arbitrage (Pittsburgh Industrial Ecosystem vs. Silicon Valley Bubble).

CONVICTION DETECTED

• That is our initiative and our goal as a company to build robotic solutions towards that. • That's what's going to happen. • We're going to be the the company that builds robots um to both identify and then solve for the most important and highest ROI problems • That's exactly the roadmap for us

HEDGE DETECTED

• I take maybe take like more like the three-year kind of time frame • You're going to need humans in the loop for some time to come • It might be nice to not have a human risking their life