The KraneShares Global Humanoid Robotics and Physical AI Index ETF (NASDAQ:KOID) was the first mainstream way to buy the humanoid theme in a single ticker. KOID has grown into a $6.3 billion fund on that pitch, carries a 0.69% net expense ratio, and is up 22.11% year to date. A second robot ETF now exists that holds a materially different portfolio, and for an investor whose thesis is specifically humanoids, KOID may be the wrong tool.
What KOID Actually Owns, and Why It Matters
A broader “physical AI” index is what KOID tracks, and its top holdings lean toward diversified suppliers rather than pure-play robotics names. Hexagon AB sits at 2.59%, Horizon Robotics at 2.45%, Magna International at 2.34%, and TE Connectivity at 2.26%. Those are robotics-adjacent businesses, though Magna is really an auto parts conglomerate and TE Connectivity is a broad-based connector maker. Owning KOID means owning the theme diluted through industrial automation and auto supply. That works as a feature for investors seeking a wider net, but it becomes a drag for those who bought a robot ETF expecting the specific mass-production ramp behind Tesla’s Optimus and Hyundai’s humanoids, which dominated headlines this summer.
The Pure-Play Alternative: HUMN
The Roundhill Humanoid Robotics ETF (CBOE:HUMN) launched June 26, 2025 and is built around who actually makes the robot. Its top holdings are Tesla at 8.93% of net assets, UBTech Robotics at 6.42%, and NVIDIA at 4.84%. Then it goes where KOID does not: Harmonic Drive Systems at 3.24%, Nabtesco at 2.20%, and Hiwin Technologies at 1.82%. These companies manufacture the strain-wave gears, precision reducers, and ball screws that make a robot arm bend without slop. There is no humanoid supply chain without them.
HUMN also carries Hyundai Motor at 2.51%, Hyundai Glovis at 2.80%, giving direct exposure to the parent of Boston Dynamics, Rainbow Robotics at 4.14%, Doosan Robotics at 2.11%, and Shenzhen Dobot at 4.15%. Roughly a third of the fund is allocated to Asian pure-plays that either build humanoid platforms or supply their joints.
Why the Composition Difference Is the Whole Argument
Tesla filed its Q2 2026 report on July 22, 2026, with $28.24 billion in revenue, up 25.5% year over year, and disclosed that first-generation Optimus production lines are being installed at Fremont on the decommissioned Model S and X lines. NVIDIA reported $81.6 billion in Q1 FY2027 revenue, with data center up 92% year over year, and CEO Jensen Huang described “the buildout of AI factories” as the largest infrastructure expansion in history. HUMN carries nearly 14% direct exposure to those two names combined. KOID’s top-10 lists neither at that weight.
Joint-builders anchor the second half of the case. If Optimus production ramps into 2026 as guided, the components that never appear in a news headline become the choke point. HUMN owns them at meaningful weights. KOID’s disclosed top-10 does not.
The Tradeoffs Are Real
Small is what HUMN is right now. Net assets sit at $46.5 million as of the March 31, 2026 NPORT filing, compared with KOID’s multi-billion-dollar base, which means wider bid-ask spreads and more single-holder risk. The expense ratio for HUMN was not disclosed in the fund’s most recently available filing, so a like-for-like fee comparison against KOID’s 0.69% net cannot be made with any confidence.
The broader mandate behind KOID will be delivered in 2026. The fund is up 22.11% year to date, well ahead of HUMN’s 8.87%, and that performance has been supported in part by the broader industrial rally, including ABB, which has climbed 39.34% year to date. Over a one-year window, the gap narrows somewhat but still favors KOID, with HUMN returning 26.36% against KOID’s 43.28%. So far, KOID’s diversification has clearly been the winning approach.
Prediction markets are skeptical about near-term Optimus milestones: Polymarket assigns only a 14.5% probability to a Tesla Optimus release by the end of 2026. Owning HUMN concentrates exposure to that timeline risk.
How to Think About the Switch
The choice comes down to which fund matches the reason a position was opened. If a KOID holder bought the fund to own industrial automation and physical AI broadly, KOID is doing that job, and the swap is not warranted. If the position was opened because of the humanoid mass-production narrative, HUMN maps to that thesis with far higher fidelity. KOID’s own returns reflect the broader basket driving performance rather than pure-play humanoid exposure. A partial reallocation, sized to the strength of the humanoid conviction and mindful of capital-gains consequences in a taxable account, respects both.
Where This Leaves the Decision
For the broader physical AI supercycle, KOID remains a reasonable core position. HUMN, on the other hand, is the humanoid ETF that the summer 2026 headlines have been describing, though it carries the concentration risk and expense disclosure gap that tend to come with being new and small. The question worth asking is whether the exposure within the ticker actually matches the story that led to the purchase, and, if not, whether trimming into a more targeted vehicle is worth the trade-off in size and fee transparency.
Contact [email protected] for any questions or corrections.
Top Gaining Stocks
Top Losing Stocks