Belridge Partners · Robotics Investment Thesis · 2026

THE
ROBOT
DECADE

Robotics companies will soon reshape US manufacturing. A step-change in human productivity is coming as the technology has arrived.

Read the Thesis
For Accredited Investors Only · 2026
Scroll
01 Investment Case

Four forces are
converging at once

A once-in-a-generation confluence of labor scarcity, technological breakthrough, and industrial policy is creating the conditions for mass robotics adoption in US manufacturing.

🧠

Tech Inflection

AI + reinforcement learning have finally unlocked dexterous, adaptable robots capable of real-world factory tasks. Physical AI has arrived.

📈

Market Scale

Global robotics market growing from $55B today to $258B by 2035 a 17%+ CAGR. Industrial robotics deal value in H1 2025 alone: $7.3B, already surpassing all of 2024.

🏭

Labor Shortage

~450K manufacturing jobs unfilled today. 1.9M gap projected by 2033. This is a structural workforce issue not a wage problem.

🇺🇸

Reshoring Surge

$1.2T in US production investments announced in 2025. CHIPS Act, IRA, tariff policy. New factories need workers that don't exist.

02 Technology Arc

From Automation
to Autonomy

Robotics has crossed the threshold from scripted motion to adaptive intelligence

01
Era 1 · 1970s–2000s
Fixed

Fixed Automation

Caged robots on assembly lines. Single task, single speed. Cannot adapt. Require complete reprogramming for new tasks.

02
Era 2 · 2005–2018
Programmable

Programmable Robotics

Vision-guided systems. Some flexibility. Cobots emerge. But still brittle unable to generalize beyond programmed scenarios.

03
Era 3 · 2019–2023
Learning

Learning Systems

Reinforcement learning. Simulation training. Robots start learning from experience. Breakout applications in logistics.

04
Era 4 · 2024–Now ▲
We Are Here

Physical AI

Foundation models for robots. Natural language control. General-purpose humanoids. Adaptive, dexterous, deployable at scale.

03 The Technology

The Technology
Has Arrived

Five decades of incremental progress gave way to a step-change. The enabling technologies converged.

AI Core

Reinforcement Learning

Robots now train in simulation and transfer skills to physical hardware. Tasks that once required months of programming are learned in hours.

Breakthrough

Physical AI & Foundation Models

GPT-style models adapted for robotics (NVIDIA Isaac, Google RT-2) allow robots to understand natural language instructions and generalize across novel tasks without bespoke reprogramming.

Mature

Computer Vision & Sensing

Multi-modal sensor fusion LiDAR, depth cameras, tactile sensors gives robots rich spatial awareness. Quality inspection that once required human eyes now runs at machine speed.

Economics

Cost Curve Collapse

Industrial robot cost per unit fell from ~$27,000 in 2017 to under $11,000 in 2025. Robot-as-a-Service (RaaS) models eliminate CapEx entirely for SME manufacturers.

Deployment

Collaborative Robots (Cobots)

Safe enough to operate alongside humans without caging. Force-limiting systems and stop-on-contact make cobots deployable in mixed human-robot environments.

Scale

Digital Twins & Simulation

NVIDIA Omniverse, Caterpillar, Boeing companies now build virtual factories to train robots before a single physical unit is deployed. Dramatically cuts implementation risk and cost.

04 Market Opportunity

Global Market
Scale

Global Robotics Market Size (USD Billions)
$55B
2025
$80B
2027
$120B
2030
$175B
2033
$258B
2035
17% CAGR
4.28M
Robots operating globally in factories today (IFR 2024)
$7.3B
Robotics deal value in H1 2025 alone
~13M
Robots in circulation projected by 2030
$24.5B
Robotics software TAM by 2030

Industrial robotics deal investment in H1 2025 surpassed all of 2022 the prior record year. The window is open.

Sources
  • International Federation of Robotics (IFR), World Robotics 2024
  • Boston Consulting Group, Global Robotics Market Forecast 2025
  • PitchBook, Robotics Deal Activity H1 2025
  • McKinsey Global Institute, The Future of Work 2025
05 The Problem

The Labor
Shortage

This is not a wage problem. It's a structural workforce collapse.

450KJobs unfilled today (BLS, 2025)
1.9MProjected gap by 2033 (Deloitte)
65%Of manufacturers cite talent as #1 challenge
10yrManufacturing lost a decade of workforce
Hardest-to-Fill Roles

Industrial Machinery Mechanics

Requires years of hands-on training; aging workforce retiring faster than replacements emerge

Quality Control Inspectors

Increasing complexity of products requires advanced technical knowledge

CNC Operators & Machinists

Highly specialized precision skills; declined school enrollment in trades

Assembly Line Technicians

High turnover, repetitive injury risk; younger workers avoid the roles

Welders & Metal Fabricators

Physically demanding; 400K+ projected shortage; 50% of workforce near retirement

Logistics & Material Handlers

Warehouse and plant-floor movers physically demanding, high injury rates

06 The Catalyst

The Reshoring
Surge

The Factory is Coming Home.

A confluence of forces is driving a historic shift back to domestic production CHIPS Act, IRA, tariff policy, and supply chain de-risking have unlocked over $1.2 trillion in announced US production investments in 2025 alone.

💡

CHIPS Act: $280B+ in US semiconductor manufacturing

IRA: $369B in clean energy solar, EV, battery factories

🏗

Tariff policy pushing electronics & auto assembly onshore

🔗

Post-COVID supply chain resilience mandates domestic production

The Math Doesn't Work

New factories are being built but there's no one to staff them.

3.2M
1.4M
1.9M

Without automation, this gap cannot be closed. The jobs exist. The workers do not.

07 Investment Framework

Two Bets on
How Robots Win

The Brain Software Platform

Hardware-Agnostic AI

A universal AI model that runs on any robot hardware, regardless of form factor or manufacturer.

Example: Skild AI
  • Hardware-Agnostic "Omnibodied"Works across arms, humanoids, mobile robots. One model, every body. Adds sensor noise in training to force generalization.
  • Internet-Scale TrainingTrains on simulation (NVIDIA Isaac) + web video to build generalized world understanding without massive teleoperation datasets.
  • Two-System ArchitectureLarge model handles high-level planning; a small, efficient model runs real-time motor control splitting cognition from execution.
  • Winner-Take-Most PotentialIf the "brain" becomes the OS layer, it captures value across every hardware platform like Android for robots.
Relies on sim-to-real transfer holding up at scale; manipulation still unproven
VS
Vertical Integration Full Stack

Own the Whole Stack

Company owns hardware, software, and data pipeline. Tight co-optimization across every layer of the stack.

Example: Tesla Optimus
  • Hardware-Software Co-DesignFull control from actuators to AI means every layer is optimized together no interface friction.
  • Proprietary Data FlywheelTesla's fleet generates unique real-world training data unavailable to software-only players. Scale compounds advantage.
  • Manufacturing ScaleTesla's manufacturing muscle means potential for cost curves inaccessible to startups. Cost per unit can fall dramatically at volume.
  • Single Deployment ContextOptimized for specific use cases factory floor, warehouse where specialization beats generalization on performance.
Capital-intensive; hardware iteration cycles are slow; single-use case limits TAM
The Window Is Now

Three Forces Converging
in Real Time

Late-stage private robotics companies are capturing this tailwind today. This SPV offers investors pre-IPO access to the defining industrial story of the decade.

🤖

The technology has crossed the threshold AI-native robots are production-ready

⚠️

The labor shortage is structural and worsening wages alone will not solve it

🏭

Reshoring is injecting $1.2T into new US factories that need robotic workforces

Inquire

Interested in
working with us?

We work with a focused group of family offices globally and high-net worth investors who share our long-term orientation. If you're a founder, co-investor, or qualified investor who believes the world is shifting rapidly, we'd like to hear from you.

All inquiries are kept strictly confidential.
We typically respond within 48 hours.