I’ve been a trader and investor for 44 years. I left Wall Street long ago—-once I understood that their obsolete advice is designed to profit them, not you.
Today, my firm manages around $5 billion in ETFs, and I don’t answer to anybody. I tell the truth because trying to fool investors doesn’t help them, or me.
In Daily H.E.A.T. , I show you how to Hedge against disaster, find your Edge, exploit Asymmetric opportunities, and ride major Themes before Wall Street catches on.
Table of Contents
H.E.A.T.
|
28–40 | 16 / 18 | 60% | $125K → $30K |
Powered joints per humanoid — each one a repeat sale | Harmonic gear score on Goldman’s 1–3 scarcity scale, the highest of any component | Discount LeaderDrive prices its harmonic gears versus Japan’s Harmonic Drive Systems | Target bill-of-materials cost-down per robot, still unproven |
Everyone is buying the robot’s face. Goldman’s scorecard says the money is in the joint.
On June 30, 2026, the Goldman Sachs Global Investment Research desk did something useful: it scored ten humanoid robot component categories on a 1-to-3 scale across content value, technology barrier, manufacturing barrier, capacity flexibility, and incremental humanoid demand. The output is a chokepoint map of an entire emerging industry. Almost nobody read it that way.
Most of the money chasing the humanoid trade is chasing the wrong floor of the building. Tesla. Figure. Unitree. The name on the chest. That is Layer 5 thinking — the same instinct that sent a crowded field piling into NVIDIA during the AI infrastructure buildout, paying a premium for the most obvious exposure to a real trend.
The bottleneck never lives where the crowd is looking. A humanoid robot is not one product. It is 28 to 40 powered joints wrapped around sensors, a power tree, and a compute stack, bolted to a frame. The company on the chest gets the magazine cover. The supplier inside every joint gets the recurring sale.
That is the Forty-Times Trade.
GOLDMAN’S SCORECARD — Ten component categories, scored 1–3 across content value, technology lock-in, manufacturing barrier, capacity flexibility, and humanoid-specific demand. The harmonic reduction gear scores highest of any category at 16. The dexterous hand follows at 15. GPU/CPU — the layer everyone is actually buying — scores 13, dragged down specifically by the two columns that matter most for a pure-play humanoid thesis: capacity flexibility and incremental humanoid opportunity. |
PART I: THE BOTTLENECK MIGRATES
Strip the skin off a humanoid robot and what is left is not a product. It is an assembly: 28 to 40 powered joints, a sensor layer, a power tree, and a compute stack, bolted to a frame. The company that bolts it together gets the magazine cover. The company that owns the part inside every joint gets the pricing power.
One joint, reduced to its components, is four stacked supplier relationships: a motor that spins, a reducer that converts that spin into slow human-grade force, an encoder and torque sensor that report position and load, and a driver chip that feeds it current. Multiply by 40. Wrap it in a chassis. That is the entire build.
The investment question collapses to one variable: which of those four layers is scarcest, and which one repeats the most times per unit. Goldman already ran that math. The answer is not the brain.
“The winner is not the company with the perfect product. It is the company that owns the part that is hardest to make and shows up the most times in every single robot.” |
PART II: WHY THE BRAIN IS THE LOW-ALPHA TRADE
Compute and foundation models are essential to a humanoid and nearly irrelevant to a humanoid-specific investment thesis. The same silicon that powers a robot also trains a chatbot, drives a car, and runs a data center. The two scoring categories that matter most for a pure-play bet — capacity flexibility and incremental humanoid demand — are exactly where the compute layer scores weakest. Humanoids are a rounding error on that demand curve. The highest-profile name in the room is, by Goldman’s own scoring framework, the lowest-alpha humanoid bet available. The exposure already sits in most portfolios through the broader AI infrastructure trade — there is no need to pay for it twice in a robot costume.
Vision and ranging — lidar, cameras — score similarly low. The category is commoditizing under Chinese price competition and shares its demand base with the automotive sector. The overlooked half of the sensor stack is the torque and tactile sensing buried inside every joint and fingertip, reporting force and grip in real time. That is a per-joint, per-robot, repeating sale — the scarce kind of demand this entire analysis is built to find.
PART III: THE REDUCER — WHERE THE TOLLBOOTH SITS
A motor alone is useless to a robot limb. It spins too fast and too weak to move anything with control. The reducer is the part that converts that raw spin into slow, precise, human-grade force — a harmonic gear for rotary joints, a planetary roller screw for linear ones. Every powered joint in a modern humanoid uses one or the other.
This is the highest-scoring component category in Goldman’s entire framework. Highest content value in the bill of materials. Highest technology barrier. Highest manufacturing barrier. And functionally, a duopoly-to-oligopoly market structure: a small number of suppliers globally with the precision manufacturing capability to produce these parts at the tolerances a robot joint demands.
Harmonic Drive Systems of Japan is the dominant global supplier — reportedly the harmonic gear supplier across Tesla’s Optimus joints. China’s LeaderDrive has built a domestic position controlling a majority share of the Chinese market and a meaningful global share, pricing its harmonic drives at roughly 40 to 60 percent of the Japanese benchmark. That is the entire bull-bear tension of this layer in one sentence: the scarcity is real, and a price war between Suzhou and Japan is the mechanism that would erode it. Watch supplier gross margins, not press releases.
THE DEXTEROUS HAND — Second-highest scoring category in the Goldman framework at 15. A single hand compresses micro-motors, miniature screws, gearboxes, encoders, and tactile sensors into a package the size of a palm — a humanoid in miniature. The risk here runs the opposite direction from the reducer: the part is valuable enough that robot makers may choose to build it in-house rather than buy it from a merchant supplier, capturing the margin themselves. The reducer bottleneck risks getting priced away. The hand bottleneck risks getting vertically integrated away. Different failure mode, same underlying lesson — know which way each chokepoint can break before sizing the position. |
PART IV: THE ASSEMBLERS AND THE TOLLBOOTH BEHIND THE TOLLBOOTH
Somewhere, a company has to take the motor, the reducer, the sensor, and the driver chip and assemble them into a tested, finished actuator. That assembly step scores nearly as high as the reducer itself — not because it is exotic, but because it carries the highest certainty of adoption of any category measured. An investor does not need to guess which reducer technology ultimately wins. The robot needs assembled actuators regardless, and the suppliers who can build them at automotive-factory scale and cost discipline — because that is where they came from — are the lower-drama way to own the motion stack. The risk concentrates instead in customer mix: when one robot program represents 40 percent of an assembler’s revenue, a single program slip is a balance sheet event.
This is Tollbooth Economics applied to a hardware supply chain rather than a financial one: find the toll collector that every unit must pass through, regardless of which brand wins the marketing war on top.
ACCESS NOTE — The thesis is cleaner than the trade execution. The purest reducer and actuator exposure sits with suppliers based in Japan and China — names that are not all easily reachable through a standard Western brokerage account. US investors may need to express this theme through ADRs where they exist, local-market access, Western suppliers with adjacent exposure, or the speculative SPAC vehicle discussed below. The bottleneck is real. Access to it is imperfect, and that gap matters as much as the thesis itself. |
PART V: THE SCARCITY VEHICLE
The thesis above argues for skipping the robot makers entirely and buying the parts beneath them. There is one structural exception, and it proves the rule rather than breaking it.
The exception is Agility Robotics. Agility is merging into the Churchill Capital Corp XI SPAC shell, with the combined company expected to begin trading under a new ticker on deal close. If the merger closes, Agility becomes the first liquid US-listed pure-play humanoid robotics company — there is currently no other direct way to express this theme in size through a single US-listed equity. Agility’s Digit unit is reportedly already deployed across multiple paying commercial customer sites in a specific, monetizable task — warehouse logistics — rather than chasing general-purpose autonomy, with several hundred million dollars in reported multi-year orders for its next-generation unit. Strategic backers reportedly include a leading AI infrastructure company, a hyperscaler, and an industrial supplier that simultaneously sits inside the reducer and linear-actuator layer of the parts map described above — meaning the same name shows up as both supplier and investor.
The case for the exception: scarcity. That scarcity matters, and it comes with real revenue and delivery dates attached — not just a pitch deck.
The case against: it is still a pre-close SPAC. Trust value sets the floor; everything above it is the market pricing in deal close and execution risk. A meaningful sponsor promote and an early lockup-release mechanism tied to a sustained stock price target both work against early holders. The order book is milestone-contingent rather than booked revenue. The entire margin thesis depends on cutting the bill of materials by roughly three-quarters — a target, not a fact, and one that runs directly into the same China-versus-Japan reducer pricing war flagged above. Treat this as speculative optionality, not the core expression of the trade.
WINNERS & LOSERS
COMPONENT / SECTOR | VERDICT | WHY IT MATTERS | RISK |
Harmonic gear / roller screw suppliers | TOP BOTTLENECK | Highest-scoring component category in Goldman’s framework. Highest content value, highest technology and manufacturing barriers, in every powered joint, 28–40 times per robot. | Chinese suppliers pricing 40–60% below the Japanese benchmark threaten to compress margins faster than volume grows |
Dexterous hand module suppliers | BENEFICIARY / WATCH | Second-highest scoring category. Highest integration complexity on the entire build, compressing an entire actuator stack into palm-sized form factor. | Robot makers may vertically integrate hand production in-house, cutting merchant suppliers out of the margin entirely |
Actuator assemblers (auto-supply-chain heritage) | BENEFICIARY | Highest certainty-of-adoption score in the framework. Robots need assembled actuators regardless of which underlying reducer technology wins; auto-supply heritage means cost and volume discipline already exist. | Customer concentration risk — a single robot program can represent a large share of assembler revenue |
Coreless / frameless motor makers | BENEFICIARY | More robots scales linearly to more motors. Genuine humanoid-specific demand and high capacity flexibility, per Goldman’s scoring. | Market is fragmented across many suppliers, capping individual pricing power versus the reducer layer |
Torque / tactile sensor makers | BENEFICIARY (overlooked half) | The sensor category scores well, but the headline vision/lidar names obscure that the better economics sit in force and torque sensing buried in every joint — a repeating, scarce sale. | Smaller addressable market than vision/lidar; less analyst coverage means less near-term re-rating catalyst |
Public automation platforms (TER, SYM, ABB, ZBRA) | ADJACENT BENEFICIARY | Not humanoid-pure plays, but these names already monetize warehouse and factory automation today, while humanoids move from demo to deployment — a more accessible Western on-ramp to the same theme. | Diluted humanoid-specific upside versus pure-play exposure; broader industrial-cycle sensitivity |
GPU / foundation model compute | ALREADY OWNED — NOT YOUR ALPHA | Essential to the robot, but humanoid demand is a rounding error against existing AI infrastructure demand. Among the lowest humanoid-specific scores on the framework despite the highest profile. | None specific to humanoids — the risk is overpaying for exposure that already exists in a typical AI infrastructure allocation |
Vision / lidar / camera suppliers | PRESSURE POINT | Among the lowest-scoring categories on the framework. Commoditizing under Chinese price competition and sharing demand with the automotive sector rather than capturing humanoid-specific pricing power. | Price war dynamics already visible; demand overlap with EV/ADAS dilutes any humanoid-specific catalyst |
Battery / cell suppliers | PRESSURE POINT | Among the lowest scores alongside vision. Cells are shared with EV demand and fully commoditized; humanoid volume is immaterial against existing battery end markets. | No humanoid-specific catalyst exists to re-rate these names independent of the broader EV/storage cycle |
Agility Robotics (CCXI → AGLT, pending close) | SPECULATIVE OPTIONALITY | Scarcity value as the only liquid pure-play humanoid exposure, with reported commercial revenue and delivery dates already in place. | Pre-close SPAC structure: trust-value floor, dilution from sponsor promote, milestone-contingent order book, unproven cost-down targets |
PRESSURE POINTS
PRESSURE POINT | WHAT TO WATCH | TIME HORIZON |
China vs. Japan reducer pricing war | Gross margin trends at harmonic gear and roller screw suppliers on both sides of the price gap — not press releases, the actual margin line | Ongoing — already underway |
Tesla Optimus 2026 production guidance | Reported guidance spans a wide unit range; the low end of that range materially cuts elasticity for the entire downstream supply chain | 2026, ongoing |
Bill-of-materials cost-down execution | Whether suppliers can hit aggressive per-unit cost targets without a margin-destroying price war across the reducer and motor layers | 2026–2028 |
Humanoid SPAC deal close | Minimum-cash redemption condition, lockup-release triggers tied to sustained share price, and sponsor promote dilution mechanics around close | 2026 |
Merchant vs. in-house dexterous hand production | Whether leading robot makers continue sourcing hand modules externally or pull the highest-margin component in-house | 2026–2027 |
CREDIBILITY FIREWALL
SOURCED / REPORTED | MODELED / INFERRED | EDITORIAL VIEW |
Goldman Sachs Global Investment Research component scorecard, 1–3 scale across content value, tech/manufacturing barriers, capacity flexibility, and humanoid-specific demand | Composite scoring weights and category boundaries reflect Goldman’s methodology, not an independently audited standard | The framework is the most rigorous sell-side attempt to rank scarcity across this supply chain that we have seen — treat the rankings as directional, not precise |
LeaderDrive harmonic gear pricing reported at roughly 40–60% of the Japanese benchmark | Margin impact on incumbent suppliers if Chinese pricing gains share is a model assumption, not a disclosed figure | This is the single largest swing factor in the entire reducer thesis — it deserves more attention than the bullish scarcity narrative alone provides |
Pending SPAC merger structure, reported commercial deployment hours, multi-year order figures, and strategic investor list as publicly reported | Bill-of-materials cost-down target from roughly $125,000 to $30,000 per unit is a stated target, not an achieved or audited figure | Scarcity value of being the only liquid pure-play is real; it does not substitute for execution risk on the cost curve |
Tesla Optimus 2026 unit guidance reported as a wide range | Supply chain revenue sensitivity to low end versus high end of that range is a derived estimate | Treat any single-program demand number across this entire supply chain as a range, not a point estimate, until guidance narrows |
BEAR CASE: WHY THIS COULD BE NOISE
Humanoid unit volumes could disappoint badly enough that even the scarcest component layer never reaches the scale needed to matter to a diversified portfolio — 28 to 40 joints per robot means nothing if total robot shipments stay in the low thousands rather than the hundreds of thousands. Recent reporting on China’s own humanoid push underscores this: heavy government and corporate investment has not yet translated into large real-world deployment, with current unit sales still small relative to the scale of the long-term narrative, even as 2035 deployment forecasts run enormous. The gap between demo and functional, paid deployment remains wide. The reducer scarcity thesis assumes pricing power holds; a China-versus-Japan price war already underway could compress margins across the entire layer before volume ever arrives to offset it. Most of the highest-conviction names in this supply chain are China A-shares that are difficult or impossible to access through a standard Western brokerage account, turning half of this map into a watchlist rather than an actionable buy list. And the single liquid pure-play vehicle remains, as of this writing, an unclosed SPAC merger — a structure with its own well-documented history of value destruction between announcement and close. |
FIVE THINGS TO DO WITH THIS INFORMATION
1. Stop pricing the humanoid trade through the robot maker’s logo. Price it through the reducer and the hand — the two highest-scoring chokepoints on Goldman’s own framework, regardless of which brand wins the marketing war.
2. Watch supplier gross margins at the harmonic gear and roller screw makers, not headlines. The China-versus-Japan pricing gap is the single mechanism most capable of eroding this entire thesis.
3. Treat actuator assemblers with automotive-supply-chain heritage as the lower-drama way to own this trade — highest certainty of adoption, but watch customer concentration closely.
4. Do not pay twice for compute exposure already sitting in a broader AI infrastructure allocation just because it is wearing a robot costume. The humanoid-specific alpha in the compute layer is, by the data, the weakest in the stack.
5. Size any pure-play humanoid SPAC exposure as a speculative sleeve with a known floor, not a core position — the scarcity of liquid access to the theme is real, but it does not erase pre-close deal risk.
The AI Buildout Has a Physical Layer
Many of today’s data centers are still using copper wiring. The same metal we’ve been using for a hundred years.
At the speeds AI demands with data moving between thousands of GPUs, billions of times a second, copper doesn’t just slow down.
It turns that data into heat. The more you push through it, the worse it gets. There’s no software for fix for that.
So what’s the answer?
Explore the Photonics Layer…..
Tuttle Capital Pure Play Photonics ETF (FOTO)
Distributor: Foreside Fund Services | Investing involves risk including possible loss of principle.
News vs. Noise: What’s Moving Markets Today
We continue to see an unwind of the AI momentum trade, but yesterday may have been a bottom, at least in the short term…..
The open was the low of the day and the close was pretty close to the high. The memory names are also up pre market.
The area is going to continue to be volatile and the SK Hynix ADR launches tomorrow. Just like the SpaceX IPO, we have never really seen anything like this before, so trying to predict what happens is just a guessing game.
News…..
China’s CXMT is starting book-building for a $4.34 billion IPO after first-quarter revenue rose 700%, while Reuters says China may let Alibaba, ByteDance, and DeepSeek buy a limited number of Nvidia H200 chips.
Fed funds futures now imply 38 basis points of tightening this year. The Fed minutes did not give the market a Powell blanket either. A few officials already saw a case to hike in June, Warsh stripped out forward guidance, and nine of 18 policymakers saw rates higher by year-end.
Bank of America raised its call to three quarter-point hikes this year. Former Fed president Jim Bullard put it plainly: "A lot of people are talking about one rate increase... The committee does not generally do that... usually it means a tightening cycle"
Noise…..
Where Does the Money Go When AI Hits a Wall?
When capital chases a tech theme, it tends to pile into the most obvious
layer and miss the one underneath. AI spending is now bumping hard
against memory. Hyperscalers — the big cloud builders like Amazon,
Google, and Microsoft — have shifted memory from 8% of their build
budgets to an estimated 30% in a single cycle. That capital has to go
somewhere. If the constraint is memory, and the build can't move without
it, shouldn't an investor own the layer AI runs on?
View HBMX fund holdings →
Distributor: Foreside Fund Services | Investing involves risk including
possible loss of principal.
<Link = http://www.hbmxetf.com/>
ETF News
A Stock I’m Watching
PENG has been represented here a few times, broke out to new highs yesterday.
In Case You Missed It
Great conversation on wide ranging topics with Kenny Polcari…
The H.E.A.T. (Hedge, Edge, Asymmetry and Theme) Formula is designed to empower investors to spot opportunities, think independently, make smarter (often contrarian) moves, and build real wealth.
The views and opinions expressed herein are those of the Chief Executive Officer and Portfolio Manager for Tuttle Capital Management (TCM) and are subject to change without notice. The data and information provided is derived from sources deemed to be reliable but we cannot guarantee its accuracy. Investing in securities is subject to risk including the possible loss of principal. Trade notifications are for informational purposes only. TCM offers fully transparent ETFs and provides trade information for all actively managed ETFs. TCM's statements are not an endorsement of any company or a recommendation to buy, sell or hold any security. Trade notification files are not provided until full trade execution at the end of a trading day. The time stamp of the email is the time of file upload and not necessarily the exact time of the trades. TCM is not a commodity trading advisor and content provided regarding commodity interests is for informational purposes only and should not be construed as a recommendation. Investment recommendations for any securities or product may be made only after a comprehensive suitability review of the investor’s financial situation.© 2026 Tuttle Capital Management, LLC (TCM). TCM is a SEC-Registered Investment Adviser. All rights reserved.