Selected track record

What prior investments taught us about the current fund.

Anduril, Figure, RADAR, Radiant, Erebor, and Harmonic AI each illuminate a different part of the Sierra Vale thesis: intelligence is becoming operational, embodied, powered, financed, verified, and deployed in mission-critical environments.

Company descriptions below are based on public information. Track-record framing reflects Sierra Vale’s investment thesis. Confirm investment attribution, permissions, and required disclosures with counsel before publication.

Filter by thesis

Each company is more than a logo. It is a learned pattern.

Filter the track record to see how historical pattern recognition informs current underwriting.

Defense autonomy · Mission systems

Anduril

DefensePhysical systemsInfrastructure layers

Public snapshot

Anduril represents the software-first rebuild of national security: autonomous systems, AI, computer vision, sensor fusion, networked defense capabilities, and rapid product iteration for U.S. and allied customers.

Why it mattered to the track record

Anduril showed that defense companies can compound like software platforms when they control the full loop: sensing, decisioning, edge autonomy, hardware, manufacturing, deployment, and operator feedback.

Fit with the current Sierra Vale thesis

For Sierra Vale, Anduril is the archetype of a defense AI company that is not merely selling a tool into the existing defense stack. It is rebuilding portions of the stack itself and turning mission urgency into product velocity.

Humanoid robotics · Embodied intelligence

Figure

Physical systemsEnterprise systemsInfrastructure layers

Public snapshot

Figure is building general-purpose humanoid robots and Helix, an AI system for humanoid control that connects perception, movement, and reasoning in real time.

Why it mattered to the track record

Figure sits at the center of physical systems. It illustrates how robotics can move from impressive demonstrations toward useful labor when hardware, software, autonomy, data collection, safety, and manufacturing are designed together.

Fit with the current Sierra Vale thesis

The current fund prioritizes founders turning frontier capability into real-world labor leverage. Figure reinforces the view that physical-systems winners own the data engine, deployment workflow, safety envelope, and customer trust needed for autonomy in dynamic environments.

Real-time retail intelligence · Physical-world data

RADAR

Physical systemsEnterprise systemsInfrastructure layers

Public snapshot

RADAR combines RFID, computer vision, sensors, software, and AI to give retailers real-time item-level visibility inside physical stores and convert store activity into actionable data.

Why it mattered to the track record

RADAR shows how the physical world becomes AI-native when previously invisible events become structured, continuous, machine-readable data. Retail is the wedge; the pattern applies across logistics, manufacturing, healthcare, warehousing, and defense operations.

Fit with the current Sierra Vale thesis

Sierra Vale views physical-world context as a core AI input. Systems that know what exists, where it is, how it moves, and what action should happen next can become the data layer for agentic operations.

Portable nuclear microreactors · Resilient power

Radiant

Infrastructure layersDefensePhysical systems

Public snapshot

Radiant is building Kaleidos, a portable nuclear microreactor designed to replace diesel generators and provide resilient power for remote villages, hospitals, datacenters, military installations, and other critical sites.

Why it mattered to the track record

Radiant expands the definition of infrastructure. As compute and autonomy move into data centers, factories, remote sites, and defense environments, power is not a background utility; it is a strategic constraint and an investable layer.

Fit with the current Sierra Vale thesis

The AI era will be bottlenecked by energy, thermal management, grid reliability, siting, and resilience. Sierra Vale’s infrastructure thesis includes the physical systems that make compute, autonomy, and mission readiness possible.

Financial infrastructure · Frontier banking

Erebor

Infrastructure layersDefenseEnterprise systems

Public snapshot

Erebor is a digital-native bank publicly described as serving technology companies in areas such as AI, defense, manufacturing, and digital assets, with a focus on modernizing financial infrastructure for frontier industries.

Why it mattered to the track record

Erebor reflects a broader Sierra Vale view: hard-technology companies need financial rails that understand procurement cycles, hardware collateral, regulated customers, unconventional assets, payroll risk, treasury requirements, and speed.

Fit with the current Sierra Vale thesis

Infrastructure is not only compute. The capital stack around frontier technology can become a strategic advantage when banking, lending, treasury, compliance, and settlement are built for the operating reality of AI, defense, hardware, and manufacturing companies.

Mathematical reasoning · Verifiable intelligence

Harmonic AI

Enterprise systemsInfrastructure layersDefense

Public snapshot

Harmonic is developing mathematical superintelligence and verifiable reasoning systems, including Aristotle, with an emphasis on formal reasoning, correctness, and reducing hallucinations in high-trust AI use cases.

Why it mattered to the track record

Harmonic points to a future where enterprise systems separate into low-stakes productivity tools and high-trust reasoning engines. In safety-critical work, “sounds right” is not enough; customers will demand proof, auditability, and correctness.

Fit with the current Sierra Vale thesis

As AI enters software verification, aerospace, finance, defense, science, infrastructure, and regulated enterprise workflows, the premium shifts from fluent generation to reliable reasoning. Harmonic maps to Sierra Vale’s thesis around verifiable AI and infrastructure for trusted autonomy.

Fit matrix

How the track record maps to Sierra Vale’s current fund thesis.

Company	Enterprise systems	Infrastructure layers	Physical systems	Defense	Primary lesson
Anduril	Mission software	Edge AI stack	Autonomous systems	Core	Full-stack systems beat legacy integration.
Figure	Labor workflows	Robot data and deployment stack	Core	Dual-use potential	Embodied AI compounds through fleet learning.
RADAR	Retail operations	Real-world event layer	Core	Operational awareness pattern	Instrumented environments create AI-native workflows.
Radiant	Critical operations	Core	Industrial deployment	Power resilience	Energy is a strategic AI and defense input.
Erebor	Regulated financial workflows	Core	Hard-tech finance	Dual-use banking rails	Frontier sectors need specialized capital infrastructure.
Harmonic AI	High-trust reasoning	Verifiable AI layer	Safety-critical autonomy	Trusted decision support	Correctness unlocks consequential AI markets.

Pattern library

Five lessons we apply to new investments.

Real-world data is the moat.

RADAR and Figure show that AI advantage compounds when products collect proprietary physical-world data competitors cannot scrape.

Infrastructure expands beyond software.

Radiant and Erebor show that compute, power, banking, underwriting, and compliance are part of the frontier stack.

Mission-critical markets reward complete systems.

Anduril demonstrates the importance of integrated hardware, software, autonomy, manufacturing, and mission understanding - a pattern that carries into physical systems and infrastructure more broadly.

Reliability becomes the enterprise wedge.

Harmonic AI points toward a future where verifiable reasoning underpins high-trust AI in safety-critical workflows.

Complex categories need clear narratives.

Each company made technical depth legible to customers, talent, strategic partners, and later-stage capital.

Public source notes used for third-party company descriptions

Anduril mission and technology overview Figure Helix overview Gap / Old Navy description of RADAR Radiant Kaleidos overview NRC Kaleidos pre-application page Reuters on Erebor banking charter reporting Harmonic official site Reuters on Harmonic funding and reasoning focus

For founders

Building inside one of these patterns?

We are especially interested in companies that combine software, atoms, mission, and infrastructure into products customers can deploy and trust.

Send us the memo