Equity is funding robot fleets.
It shouldn’t have to.

Financial infrastructure for scaling robot fleets

Building the SoFi for robots

Bob Liu · JP Daum

Overview · Confidential

The Setup

Physical AI and Robotics are coming after the $4.8-trillion U.S. labor market

$4.8 trillion
labor market

$68m physical labor jobs in the U.S.

$49k avg-weighted annual wage

$70k total comp

Persistent labor shortage

VC money is
pouring in

$7.4bn raised in 2025

$10.2bn raised in 2026 YTD

79+ vertical robotic startups raised a round in 2026

Robots priced
as OpEx

Economic value comparable to human labor

Startups keep high-margin, recurring cashflow streams

Robot-as-a-Service: “you are paying minimum wage per hour to this robot to do 3 people’s job!” Startups keep a high-margin, recurring cashflow stream.

Source: BLS (2025); Crunchbase and FleetScale analysis, as of August 2026. Labor market context — not a FleetScale book forecast.

Problem

However, RaaS creates a cashflow issue…

Customer wins

No upfront CapEx

Automation becomes an OpEx line, with efficiency gains flowing directly to the bottom line.

Adoption opens to any size company

Robotic startups carry the book

Startups become capital intensive
Every deployment ties up cash
Growth consumes the balance sheet
Scaling requires expensive venture equity — or stops

Robotic startups are getting capital intensive and using expensive equity to scale.

What we do

We sit between robotics companies and capital

Demand

Robot companies

Need fleet CapEx without burning equity

Standardized, redeployable units

Platform

FleetScale

Originate and underwrite

Pool, tranche, and syndicate

Monitor. Issue again.

(Securitization platform path)

Supply

Capital

Private credit

Specialty finance

ABS investors

Equity builds the rails. Warehouse and takeout facilities fund the fleets.

Network effect

More fleetsSites on rails

Better dataLoss · util · service

Tighter priceHonest advances

More capitalInstitutions lean in

Competitive landscape

Why not hedge funds, fintechs, lessors, or captives?

Capital is not scarce. Shared definitions, multi-vendor books, and a path to pooled paper are.

Who	What they optimize	Where they stop	FleetScale wedge
Hedge funds / credit opp	Yield on existing, diligenced assets; bilateral special sits	No origination standards for novel fleets; no desire to be the category data room	Make fleets legible so HF capital can buy senior / mezz without reinventing robotics credit
Fintechs into “new assets”	Consumer/SME rails, marketplace ABS path, software distribution	Still cash-flow paper they already model — not robot util + service correlation + redeploy recovery	Primary object: contracted robotic work + redeployable serial units, multi-OEM
VC / growth BDCs & venture debt	Company-level senior debt to equity-backed growth cos	One-off facilities; not a pooled fleet book with shared util/loss definitions	Different asset: work paper + fleet — not generic growth term loan
Equipment lessors	Forklift residual curves, serial remarket playbooks, tax/lease product	Thin residual markets for novel robots; software/service risk; fixed-cell bias	Redeploy-first recovery; residual junior; fee stream primary
OEM captives	Finance their own metal to win the hardware sale	Single-vendor book; duration on OEM balance sheet; no multi-vendor syndication	Neutral rails: underwrite and pool across builders when standards exist
Banks (later)	Warehouse/ABS once history, ratings, and definitions exist	Will not invent category standards at pre-seed	Pre-seed job: standards + underwriting surface so banks can enter later

Opinion / thesis map — not a competitor diligence report. BDC yield band is lender book context only (not our return).

Novel asset class formation

Novel asset class formation creates durable equity value

Consumer credit as an example

2011

SoFi founded

Four students. Pilot: 40 alumni lend $2M to ~100 students. No ratings, no shelf, no market.

2013

First securitization

SoFi’s inaugural ABS: $152M of senior notes, rated single-A. First public securitization by a marketplace lender.

2016

A cohort forms

$4.6B issued. Multiple originators, branded shelves, rating agencies building frameworks.

2017

The inflection

$7.8B issued, +71% YoY. Six issuers at scale. Ratings migrate single-A → AA; SoFi reaches AAA.

Today

Core fixed income

Unsecured consumer loan ABS set records in 2025 at roughly $25B.

The cohort today · FY2025 consumer unsecured marketplace-lending ABS issuance

$5.79B

$4.38B

$2.28B

Oportun

$2.00B

Upstart

$1.53B

$1.28B

One asset class produced multiple public companies (LendingClub, Upstart, Affirm, SoFi, Pagaya). Durable value accrued to the securitization layer. We are building that layer for robots. · Sources: KBRA, Cross River, SEC filings.

Why us

Cofounders background

The bridge between VC-backed robotics originators and institutional credit

Bob Liu

Institutional capital allocator

Managed an asset-based lending and specialty finance portfolio at an institutional investor.

Standing relationships with the specialty finance / credit funds that buy first-time issuers.

Coverage also included both venture capital and leveraged buyout.

JP

Structured credit, ABS, systems

Structures and executes asset-backed transactions.

Warehouse mechanics, tranching, rating-agency process.

Startup and core engineering background.

Full names, titles, and prior firms to lock before external send.

Underwriting surface

What we underwrite — specifically.

Primary credit object = auditable contracted work. Unit residual is recovery / haircut, not the advance thesis.

Primary (advance)

Billable / contracted fee stream (RaaS, service SLA, recurring work orders)
Serial identity per unit + site / fleet ID
Utilization definition (hours, cycles, jobs) agreed in data room
Obligor / operator cash-flow story independent of OEM equity raise
Service step-in rights if the builder stumbles

Support (haircut)

Redeploy path: second site / second customer before scrap
Hardware residual after explicit haircuts (never sole recovery)
Software/version risk flags (correlation with site failure)
Insurance / maintenance contracts where they bind
True-sale / SPV path as process field (not claimed closed)

Auto-no (refuse)

No auditable billable unit (residual-only advance)
Fixed cell with no exit value off one site
Bespoke system tied to one obligor, no redeploy story
Utilization that cannot be independently evidenced
Present-tense “we’re already ABS” with no book or standards

Product claim (path): definitions → data room → advance logic → monitor. Not a live portfolio checklist.

Equity is funding robot fleets.It shouldn’t have to.