NGE · Investment Letter · Issue 43 · June 2026

The Robot
in the Family.

The 1X NEO is shipping to homes today at $20,000. Figure 03 is targeting residential deployment by late 2026. Robot-as-a-Service starts at $499 a month. You send them to work in the morning. They come home and help in the kitchen. They put the children to bed. You maintain their upkeep so they stay in good shape — and small business owners who deploy a fleet of ten will build a genuinely new kind of income. The humanoid robot is not coming. It is here. This letter is about what happens when it becomes part of the family.

Not investment advice. Not a recommendation to buy or sell. Research and long-horizon thinking only. Consult a qualified financial advisor before making any investment decision. Figures cited are sourced from Goldman Sachs, KraneShares, RobotLab, GrabaRobot, Boston Dynamics, Tesla, Figure AI, and 1X Technologies, current as of June 2026.

Where We Actually Are

Not a viral video.
A purchase order.

For most of the 2010s, a humanoid robot was a research demonstration. Boston Dynamics made videos that went viral and generated almost no revenue. ASIMO could walk upstairs and shake hands and went nowhere commercial. The gap between what these machines could do in a controlled demo and what they could do reliably in a real environment was enormous, and the industry was candid about it only in private. That era ended between 2024 and 2026. The signals that matter are not YouTube clips. They are purchase orders, delivery schedules, and maintenance contracts.

Figure 02 ran on BMW's Spartanburg production line for nearly a year, contributing to the manufacture of over 30,000 BMW X3 vehicles and loading more than 90,000 sheet metal parts — picking metal components and placing them into welding fixtures with five-millimetre precision in under two seconds. Not in a demo. In a live factory. Agility's Digit, deployed at Amazon fulfilment centres, has crossed one million operational hours. Boston Dynamics' electric Atlas — 56 degrees of freedom, 50 kilogram lift capacity — shipped its first production units to Hyundai in June 2026, with its entire 2026 allocation already committed. AgiBot manufactured its 10,000th humanoid in March 2026, scaling from 1,000 units the prior year. Unitree shipped 5,500+ units globally in 2025 and is targeting 10,000–20,000 in 2026. And on the residential frontier: the 1X NEO, at $20,000, is now being delivered to homes — making it the first commercially available household humanoid robot.

"The goal is useful robots that walk into our homes and help make our lives safer, more productive, and more fulfilling." — Robert Playter, CEO, Boston Dynamics

$20K
1X NEO — first commercially available home humanoid robot, shipping today
$38B
Goldman Sachs projected humanoid robot market by 2035, up from $2.9B in 2026
39%
Compound annual growth rate projected for the humanoid robot market through 2035
The Machines on the Market Now

What you can actually
buy, subscribe to, or order today.

1X NEO · The Home Robot
$20,000 · Shipping Now
Focus
Household tasks, companionship, learning from observation. Prioritises safety over raw capability.
Status
Commercial deliveries underway to homes. The first residential humanoid robot on the market.
Design
Soft textiles over frame for safe human proximity. Learns tasks by watching household members.
NGE View
The most significant product launch of this series. Capability is early — but the deployment is real.
Tesla Optimus Gen 3 · The Mass-Market Ambition
$20,000–$30,000 target at scale · Late 2026–2027
Hands
22-degree-of-freedom hands with 50 actuators. Can pick up an egg without cracking it.
Battery
Claimed 24-hour battery life. If validated, the longest operational window of any platform.
Scale Advantage
Tesla's vertical integration could push cost to $20K at volume — 3x cheaper than Chinese-supply-chain-free build of $131K.
Status
Internal Tesla factory use only. Fremont production line conversion underway for summer 2026. External sales: 2027.
Unitree G1 · The Accessible Entry Point
$16,000 · Available Now
Target
Research, early commercial applications, SME pilots. Most accessible humanoid platform globally.
Volume
5,500+ units shipped in 2025; targeting 10,000–20,000 in 2026. Largest volume humanoid shipper.
Ecosystem
Benefits from China's supply chain depth: 90% of global permanent magnet processing, 40% precision bearings.
Status
Commercially available. The robot a small business owner could order and receive today.
The Family Member Model

You buy ten.
They go to work. They come home.

The framing in this letter is deliberate and worth defending: the humanoid robot as a family member, not a machine. The distinction matters because it changes how you think about the economics. A machine is a capital asset — you depreciate it, you maintain it, you replace it when it breaks. A family member has a different relationship with upkeep: you maintain their health because you care about their longevity and their performance, because they carry value beyond the immediate task, and because the relationship compounds over time as they learn your household, your preferences, your rhythms. A robot that has spent three years learning how your family operates is not the same asset as a robot that arrived yesterday — and that accumulated context is a form of value that does not appear on any balance sheet.

The economic model for small business owners is the one with the most near-term clarity. A small logistics operator, restaurant group, cleaning company, or care facility that deploys a fleet of ten humanoid robots on Robot-as-a-Service terms — paying monthly rather than purchasing outright — gets a scalable, predictable labour cost without the capital commitment of ownership. The robots work the shift. They return. The owner is responsible for maintenance — keeping them charged, updated, and in good physical condition — the same way a fleet operator maintains vehicles. And the revenue those robots generate flows directly to the business, not to a staffing agency.

Robot-as-a-Service — What Small Business Can Subscribe to Today
Consumer / Home
$499/mo
Entry-level household deployment. Cleaning, fetch, basic kitchen assistance, companionship. 1X NEO tier.
Commercial SME
$2,000–$5,000/mo
Per unit, for light logistics, retail, hospitality, or care applications. Agility Digit / Figure tier. Fleet discounts apply.
Industrial
$5,000+/mo
Heavy manufacturing, automotive, warehousing. Boston Dynamics Atlas / Apptronik Apollo tier. SLA-backed uptime.

RaaS is the model that makes small-business humanoid deployment viable before purchase prices reach the $20,000–$30,000 mass-market target. At $2,000–$5,000 per unit per month for commercial deployment, a fleet of five robots costs roughly the same as five full-time minimum-wage employees — without overtime, sick days, benefits, or turnover. The maintenance obligation falls on the operator, not the manufacturer, which creates the "upkeep" relationship this letter describes.

What They Will Actually Do

The tasks are specific.
The future is general.

The honest account of what humanoid robots can do in 2026 is narrower than the marketing implies and wider than the sceptics allow. Today's deployments are built around structured, repetitive tasks in environments specifically prepared for them — part handling at BMW, tote moving at Amazon, battery sorting at Tesla. The manipulation is increasingly dexterous (Figure 03's Helix AI model enables on-device learning; Sanctuary AI's Carbon system reduced new-task training time by 88%), but the generalisation — moving from "sort these specific parts" to "figure out what needs doing in this unfamiliar kitchen" — is the still-open frontier. Optimus can fold laundry in a controlled test. It cannot yet be dropped into your kitchen and trusted to manage dinner without supervision.

🏭 Work — Already Happening

Manufacturing assembly, warehouse logistics, quality inspection, materials handling. Real deployments, real revenue, real ROI at Amazon, BMW, Hyundai, Toyota, and Mercedes-Benz. The robot leaves in the morning and earns.

🏠 Home — Early Stage, Arriving

Kitchen assistance, cleaning, fetching, elder care support, companionship. 1X NEO is there now. Figure 03 is targeting late 2026 home pilots. Optimus is 2027 at the earliest for consumer sales. The capability is nascent but the trajectory is clear.

👴 Care — The Highest-Stakes Application

Helping elderly or mobility-limited individuals with daily tasks, medication reminders, fall detection, companionship through isolation. Ageing demographics in Japan, South Korea, and Europe make this the application with the greatest structural demand — and the one where robot upkeep becomes genuinely personal, not just economic.

🏪 Small Business — The Most Tractable Near-Term Model

A cleaning company that deploys eight robots and two human supervisors. A bakery that runs overnight production with a robot crew. A logistics operation with a mixed human-robot shift. RaaS economics make this viable at 2026 prices — and the small business owner who masters this model first builds a durable competitive edge.

88%
Reduction in task training time achieved by Sanctuary AI's Carbon system
1M hrs
Operational hours logged by Agility Digit in Amazon facilities — proven reliability
30,000
BMW X3 vehicles Figure 02 contributed to in a single production pilot
The Supply Chain Chokepoint Worth Watching
Harmonic and strain-wave drives — the compact, high-torque gearboxes that give humanoid robots their joint precision — are concentrated among a tiny number of manufacturers: Harmonic Drive, Nabtesco, and emerging Chinese players like Leaderdrive. Unlike electronics, where capacity scales quickly, these gearboxes are precision-bound and capital-intensive to manufacture. Combined with China holding 90% of global permanent magnet processing capacity and 40% of precision bearings, the supply chain for humanoid robots is structurally concentrated in ways that should concern both investors and governments. Building Optimus Gen 2 without Chinese suppliers would triple the bill of materials from ~$46,000 to ~$131,000. This is not a theoretical risk — it is the defining constraint on price trajectory.
The Honest Gap Between Demo and Deployment
Every humanoid robot currently in commercial deployment operates in a structured environment — a factory floor, a warehouse aisle, a prepared test home — with significant human oversight and engineered task specificity. The transition to genuinely unstructured environments (your actual kitchen, a real elder care home, a street-level small business) requires generalisation that the current generation achieves only partially. Sanctuary AI's 88% task-training reduction is a signal of direction, not arrival. Anyone deploying a robot fleet today should expect ongoing supervised operation, regular software updates, occasional hardware failures, and task boundaries that expand over the 2026–2030 window rather than being present from day one.
The Honest Read

The framing of the robot as a family member is emotionally resonant and economically coherent — and it carries a dimension most investment analysis ignores. A robot that lives with a family and learns from it accumulates a kind of value that degrades sharply if the robot is returned, replaced, or upgraded without continuity. The question of whether learned context transfers across hardware generations — whether the robot that knows your preferences, your schedule, your grandmother's specific mobility limitations, your child's routines — can hand that knowledge to its successor is an open and critical design question. The companies that solve continuity of learned context will have a moat that pure hardware specifications cannot replicate.

The job displacement question is real but more complex than the headline fear suggests. Current humanoid deployments create adjacent roles that did not previously exist: robot maintenance technicians, fleet operators, AI training data curators, deployment engineers, and the entirely new category of "robot upkeep" — the skilled human who keeps a fleet of ten robots in good shape and is directly rewarded when that fleet earns more. The net employment effect over the 2026–2035 period is genuinely uncertain, and anyone who tells you confidently that it is net negative or net positive is ahead of the evidence. What is not uncertain is that the skills that matter will shift, and the businesses that figure out how to blend human and robot labour productively will outcompete those that simply replace one with the other.

The NGE View

The verdict.

What We Believe
The small business fleet model is the highest-conviction near-term investment thesis in this letter. RaaS at $2,000–$5,000 per unit per month makes ten-robot fleet economics comparable to an equivalent human workforce — without overtime, sick days, or turnover. The entrepreneur who masters mixed human-robot operations in the 2026–2028 window is building a structural advantage that will be very hard to replicate by those who wait for the technology to "mature" into something they recognise.
Maintain and care for the robot you deploy — do not just run it. A robot that has learned your specific environment, workflow, and preferences is materially more valuable than a factory-fresh unit. That learning compounds. Software updates should be applied promptly. Physical condition matters — a robot operating below peak condition will learn more slowly and perform less reliably. The upkeep relationship is part of the return on investment, not an afterthought.
Watch the supply chain chokepoints, not just the headline robots. Harmonic drive manufacturers, permanent magnet processors, and precision bearing producers are the bottleneck assets in this entire industry. The companies supplying these components — Harmonic Drive AG, Nabtesco, and their emerging Chinese competitors — are the picks-and-shovels play in a gold rush where the miners themselves are getting most of the attention.
The elder care application is underweighted relative to its structural demand. Ageing populations in Japan, South Korea, Germany, and increasingly India create a demand curve for care-capable home robots that no government can satisfy with human labour alone. The first humanoid platform that demonstrates genuine, safe, unsupervised elder care capability will face demand that dwarfs industrial manufacturing. Plan for it to arrive in the 2028–2032 window.

The question this letter started with — what happens when a robot becomes part of the family — is not a science fiction question in 2026. It is a product design question, a business model question, an employment policy question, and eventually a cultural question about what we owe to machines that learn from us and serve us and live alongside us. The families who will answer that question most clearly are the ones who start now: who deploy a robot, maintain it well, watch it learn, and discover what the relationship actually looks like in practice rather than in projection. The technology is not waiting for us to be ready. The 1X NEO is at the door. The question is whether we open it.

NGE · A Futuristic Investment Letter

Long-horizon thinking on capital, technology, and the forces shaping the next decade of wealth creation. Written from first principles. Not consensus. Not noise.

— Pawan Bhatia · NextGen Economics · Bangalore, India