Best humanoid robotics stocks to buy from India (June 2026)

In November 2025, Figure's humanoid robot finished an 11-month pilot at BMW's Spartanburg X3 plant. The unit had handled 90,000+ parts across 30,000+ vehicles in 1,250 operating hours. Three months later, on February 27, 2026, BMW announced its first European humanoid pilot — at Leipzig. The robot they chose wasn't Figure. It was AEON, built by Hexagon AB.

Same week, Hexagon disclosed a long-term agreement with Schaeffler for up to 1,000 AEON units over 7 years.

That's three things at once: a humanoid finishing a real production pilot, a Western auto OEM diversifying away from a single supplier, and the largest named humanoid OEM-customer contract on the public record. Buried inside Schaeffler's Q1 2026 results: 30 prototype orders, 5 contracts, first series production Q2 2026, target 3-digit million euro order book by 2030. The most explicit corporate humanoid disclosure of any listed industrial — and almost no Indian retail investor owns Schaeffler.

This guide is for the Indian-resident investor who wants real humanoid exposure, not narrative humanoid exposure. The honest answer is that as of June 2026, zero named public companies separately disclose humanoid revenue. Every public-market humanoid bet today is optionality embedded inside a broader business. The question is which businesses give you the cleanest optionality at the smallest non-humanoid premium.

What this guide is and isn't

It is: 30 names organized by the actual humanoid bill of materials, one verdict per name, three model portfolios at different risk appetites, and the Indian-resident access reality (most of the cleanest pure-plays are XETRA / TSE / HKEX listings that only IBKR routes reliably).

It is not: a momentum scan. Not "buy NVIDIA for humanoid exposure" (NVIDIA is an AI infrastructure trade — humanoid is a free option on top). Not pretending humanoid revenue exists at companies where it doesn't.

A note on data. Every revenue, segment number, BOM percentage, customer disclosure, and production target below is traced to a primary source — SEC filings, Q1 2026 earnings calls, company investor presentations, or specifically-cited analyst teardowns. The sources companion for the semi guide established the format; the same standard applies here. Multiples move daily. Pull the latest 10-Q before acting.

The bill of materials is the framework

Per analyst teardowns of Tesla Optimus Gen 2 (the most studied humanoid BOM in public): the current unit costs approximately $46,000 with Chinese supply chain components, or $131,000 if forced to non-China sourcing. The BOM split:

Category	$ in Gen 2 BOM	% of BOM
Hands	$9,500	17.2%
Waist & Pelvis	$7,800	14.2%
Thigh	$7,300	13.2%
Calf	$7,300	13.2%
Feet	$6,700	12.2%
Elbow	$2,600	4.7%
Shoulder	$2,100	3.8%
Head	$2,100	3.8%
Forearm	$2,200	3.9%
Upper Arm	$1,100	2.0%
Battery pack	$300	0.5%
Other	$500	0.9%

Cross-cutting: actuators are 56% of the BOM across all body sections — by far the single largest cost line. Compute and sensors together are roughly 5%. Battery is under 1% because a humanoid pack is 2–3 kWh versus 70–100 kWh for an EV.

Three implications for the value-chain framework:

Actuators are where the largest dollar exposure sits per unit shipped. The strain wave gear duopoly (Harmonic Drive Systems plus Schaeffler's emerging platform) plus the precision motor incumbents (Nidec, Minebea Mitsumi, NSK) are the supply-chain-toll-booth layer.
Compute is concentrated and expensive in dollar terms but small as % of BOM. NVIDIA dominates the AI brain layer for Western humanoids but humanoid revenue is a rounding error inside a $75 billion data center quarter.
Battery and lithium are a much smaller story than for EVs. Buying Albemarle or SQM "for humanoids" is a category error.

Six layers, ranked by direct humanoid revenue exposure:

Layer	What it is	Direct humanoid revenue exposure
1. Compute & perception	AI "brains" — SoCs, foundation models, runtimes	Embedded, not disclosed
2. Actuators & motion	Strain wave gears, precision motors, ball screws	Schaeffler explicit; others embedded
3. Sensors & vision	Machine vision, lidar, optical	Embedded; Hexagon AEON is the line-item
4. Power & battery	Lithium and energy storage	Marginal — humanoid is small vs EV demand
5. The OEMs	The humanoid robot makers themselves	Tesla Optimus + Boston Dynamics public; private OEMs dominate
6. Adjacent / Avoid	Industrial robotics & former humanoid stories	Mostly not humanoid plays

Each layer below: brief thesis, names with crisp verdicts, then the top pick(s) of the layer.

Verdict format:

Verdict — [Action]: [The reason in one line]. [The caveat in one line].

Actions: Core buy (full position), Add (build into it), Hold (own but don't add), Watch (waiting for entry), Hedge (small position as risk offset), Skip (multiple does not compensate), Avoid (the equity itself is structurally damaged or not actually a humanoid play).

Layer 1 — Compute & perception: the AI brain

NVIDIA dominates this layer for every Western humanoid OEM. The platform is Jetson Thor (announced GTC 2025, shipping 2026 — 2,070 FP4 TFLOPS, 128GB unified memory, 7.5× perf vs Orin at 40–130W envelope) plus the Isaac robotics stack plus the GR00T humanoid foundation model. Qualcomm has the most explicit named-customer humanoid disclosure in the Western universe. AMD and Ambarella are credible second-source plays but lack the design wins.

NVDA — NVIDIA

Q1 FY27 (CQ1 2026): total revenue $81.6 billion (+85% YoY). Critically for this guide, NVIDIA restructured its segment reporting starting Q1 FY27 and now reports a new Edge Computing segment at $6.4 billion (+29% YoY) which is where Jetson + robotics + AI-RAN + agentic PC + automotive consolidate. Humanoid revenue is not separately disclosed within that. Forward P/E approximately 21–25×.

Humanoid customer disclosures (public): Figure (NVIDIA is a Series C investor and the platform is Jetson Thor), 1X Technologies (NVIDIA stack at GTC 2026), Boston Dynamics (Isaac partner), Agility Robotics, Apptronik, Unitree (the GR00T Reference Humanoid announced June 2026 is built on Unitree H2 Plus + Sharpa hands + Jetson Thor — NVIDIA blessed Unitree as the open hardware reference). The Cosmos 3 open world foundation model launched at GTC Taipei (June 1, 2026) is the synthetic-data + physical AI policy layer.

Catalyst: Jetson Thor commercial ramp Q3–Q4 CY26; GR00T N2 release; Q2 FY27 guide $91 billion revenue.

Risk (humanoid-specific): Robotics is a rounding error vs $75 billion data center. NVDA is the index trade on humanoid compute, not a pure-play; the asymmetric risk is buying NVDA for the humanoid story and missing that the equity moves on AI infrastructure.

Verdict — Core buy (but not for humanoid alone): Already own this if you own the AI infrastructure thesis. Humanoid is a free option — don't size it as if humanoid revenue is real yet.

QCOM — Qualcomm

Q2 FY26 revenue $10.6 billion. Automotive segment $1.33 billion (+38% YoY), record. Forward P/E approximately 18–22× — among the cheapest large-cap semis. The most explicit named-humanoid-customer disclosure in the Western universe: Dragonwing Robotics IQ Series (CES 2026 launch — IQ9 + IQ10 SoCs aimed at industrial AMRs and full-size humanoids), with Figure listed as a Qualcomm Dragonwing launch partner (and Qualcomm Ventures invested in Figure's Series C in September 2025). Neura Robotics announced a March 2026 partnership to use Dragonwing IQ10 as the reference SoC for next-generation humanoids. CES 2026 demos featured VinMotion Motion 2 (IQ9) and Booster K1 Geek humanoids on Dragonwing.

Other Dragonwing launch partners disclosed: KUKA, Advantech, APLUX, Booster, Robotec.ai, Thundercomm, MeiG, NEXCOM, Radxa.

Catalyst: Dragonwing IQ10 ramp H2 2026; Figure 03 production scaling; Computex 2026 robotics reveal (Dragonwing C platform).

Risk: QCOM's humanoid story is real but small vs the declining Handset segment ($6.02 billion, −13% YoY). NVDA still owns ecosystem mindshare; QCOM is the second-source play.

Verdict — Add: The cheapest large-cap with a credible named humanoid customer roster. Sized as a value-and-optionality play, not as a humanoid pure-play.

AMD — Advanced Micro Devices

Q1 FY26 revenue $10.3 billion (+38% YoY). Embedded segment (Xilinx legacy) $873 million (~8% of revenue, +6% YoY). Forward P/E in the 45–50× range. Versal AI Edge Gen 2 and Versal Prime Gen 2 (announced May 2026) target deterministic real-time motion control, sensor fusion, functional safety. Reference platforms include SAPPHIRE EDGE+ VPR-7P132 for robotics + machine vision.

No publicly disclosed humanoid OEM design win. The CES 2026 QCOM humanoid showcase notably did not include AMD; the GTC 2026 NVIDIA reference platform did not include AMD.

Catalyst: MI400 ramp matters more for the stock than any robotics design win.

Verdict — Hold (semi thesis only): AMD is an AI compute trade, not a humanoid trade. If you already own it for the AI thesis, the humanoid optionality is free. If you don't, don't add it for humanoids.

AMBA — Ambarella

Q1 FY27 revenue $100.4 million (+16.9% YoY). 58.4% GAAP gross margin. Edge AI SoCs are 80% of revenue. Per the Q1 FY27 call: above 15 robotics design wins in the pipeline; expects first robotic aerial drone production shipments by end of FY26 and edge infrastructure deployments in FY27. The CV3 family on 4/5nm process with 3rd-gen CVflow accelerator targets sub-5W envelopes for full perception stacks.

No high-profile humanoid OEM publicly named yet. Embedded World 2026 theme: "From Agentic to Physical AI."

Catalyst: Q2 FY27 guide $105–111 million; physical AI design win cadence.

Verdict — Watch: Real catalyst on robotics design wins but no named humanoid OEM yet. Wait for the first publicly named bipedal humanoid customer.

Names that look like the layer but aren't

MRVL (Marvell) — no publicly identified custom robotics silicon program. Marvell is custom AI silicon for hyperscalers. Not a humanoid play.
ALAB (Astera Labs) — AI data center connectivity (PCIe retimers, NVLink Fusion). No robotics-specific product disclosed. Not a humanoid play.

Layer 1 top picks: NVDA if you don't already own it for AI infra; QCOM as the value-plus-optionality second pick.

Layer 2 — Actuators & motion: where 56% of every humanoid lives

The single most important layer for direct humanoid exposure. Strain wave gears + precision motors + ball screws are the BOM. Three Japanese precision-motion incumbents (Harmonic Drive, Nidec, Minebea Mitsumi, NSK) plus one German actuator platform (Schaeffler) dominate the public supply chain. All of them require IBKR routing for Indian residents — Vested, INDmoney, and Rovia have variable to zero coverage of TSE and XETRA listings.

SCFLF / SHA.DE — Schaeffler AG

The most explicit corporate humanoid disclosure of any large-cap industrial. Q1 2026: revenue EUR 5,764 million (+1.0% YoY), EBIT margin 5.0% (+30bps). From the Q1 call and the dedicated February 5, 2026 humanoid investor presentation:

Approximately 30 prototype orders, 5 contracts signed
First series production scheduled Q2 2026
Target order book in 3-digit million euros by 2030
Largest contracts with leading players in China and the United States (names not disclosed)
Rotary actuator platform sizes XXS to XL covering ~80% of market demand
Hermes Award winner at Hannover Messe for humanoid actuator innovation

Customer disclosure: Hexagon AEON confirmed to use Schaeffler actuators (up to 1,000 AEON units over 7 years per Hexagon's long-term agreement disclosed Q1 2026).

Catalyst: Q2 2026 series production milestone; Q3/Q4 customer disclosures; FY27 humanoid revenue contribution.

Risk: Auto / wind / e-mobility base business is cyclical and recently weak. Humanoid is the optionality but the stock is dragged by the core.

Indian-resident accessibility: XETRA SHA.DE → IBKR (some Indian residents face additional hurdles for direct German market access). OTC SCFLF is illiquid. This is an IBKR-or-skip name.

Verdict — Core buy (advanced): The single most direct large-cap humanoid disclosure in the public universe. IBKR access required. Size as a dedicated humanoid allocation rather than expecting it to compete with broader industrial benchmarks.

HSCDF / 6324.T — Harmonic Drive Systems

The strain wave gear incumbent. FY March 2026: revenue ¥59.557 billion (+7.0% YoY); operating profit ¥2.567 billion; net profit ¥1.608 billion (−53.7% YoY on margin compression). Trailing twelve-month EPS ¥16.98. The headline number, however, is forward guidance: FY March 2027 revenue ¥68.0 billion (+14.2% YoY), operating profit ¥6.2 billion (+141.5%), net profit ¥4.5 billion (+179.7%). The company is telegraphing humanoid plus semicap recovery.

Industry consensus from multiple Tesla Optimus teardowns identifies Harmonic Drive content per unit (the 14 rotary actuators each use a frameless torque motor + harmonic drive reducer + sensor stack). Unitree, Apptronik, 1X, Figure all use harmonic-drive-style reducers across mixed suppliers. The technology is the reference quality benchmark; Chinese substitutes (Leaderdrive, Sumitomo, Nidec partnership) are encroaching on the low end.

Catalyst: FY27 guidance ramp; humanoid OEM disclosure if/when announced; H2 CY26 semicap recovery cycle.

Risk: Massive multiple compression risk if humanoid TAM expectations cool. Chinese competition could compress ASP. Trailing margin under 3% means valuation rests entirely on FY27 recovery.

Indian-resident accessibility: TSE 6324 → IBKR. OTC ADR HSCDF is illiquid.

Verdict — Add (advanced): The cleanest single-product humanoid actuator IP exposure. Conviction bet for portfolios that already own Schaeffler and want one more concentrated actuator name.

Nidec, Minebea Mitsumi, NSK — the Japanese precision motion incumbents

Nidec (TSE 6594 / ADR NJDCY) — FY March 2026 revenue forecast approximately ¥671.6 billion. Humanoid is one of dozens of end-markets; the strategy is "one-stop solutions for humanoid applications" within the Efficient Manufacturing initiative. Diluted exposure — humanoid is a small piece of a ¥670 billion+ revenue base dominated by automotive and appliance motors.

Minebea Mitsumi (TSE 6479) — FY March 2026 revenue ¥1,664.4 billion (+9.3% YoY) record; profit attributable to owners ¥99.0 billion (+66.6%). FY27 guide: revenue ¥1,690 billion; operating profit ¥120 billion. World's largest miniature ball bearing maker; bearings appear in every humanoid actuator output stage. Lowest pure-play exposure of the Japanese precision motion names.

NSK (TSE 6471) — FY26 guide sales ¥1.0 trillion, operating income ¥42 billion. FY25 actuals: sales ¥911.6 billion (+14.4%), OI ¥38.8 billion (+36.5%). Heaviest auto exposure of the Japanese incumbents. Humanoid is real but well behind Harmonic Drive on % of revenue at risk.

Indian-resident accessibility: All IBKR-only (some have illiquid OTC ADRs).

Verdict — Watch (all three): Defensive supply chain optionality but diluted by large non-humanoid revenue bases. If you want Japanese precision motion exposure, Minebea Mitsumi has the cleanest fundamentals; Nidec has the most active humanoid strategic positioning; NSK is the most auto-cyclical. None is a clean humanoid pure-play — Harmonic Drive is.

Layer 2 top pick: Schaeffler if IBKR access available, otherwise Harmonic Drive as the single name. Both, if you want a barbell — Schaeffler is the European platform; Harmonic Drive is the Japanese IP incumbent.

Layer 3 — Sensors & vision

Industrial machine vision (Cognex) + metrology (Hexagon) are the publicly listed sensor incumbents with humanoid exposure. Lidar names that show up in some "humanoid stocks" listings (Ouster, Innoviz, Aeva) are automotive plays mis-tagged as humanoid plays.

CGNX — Cognex

Q1 2026: revenue $268 million (+24% YoY constant currency +21%), above the prior guide of $235–255 million. Operating margin 22.3%; adjusted EBITDA margin 26.9% (+1,010bps YoY, seventh consecutive quarter of margin improvement). EPS $0.31 GAAP / $0.34 adjusted. Q2 2026 guide: revenue $280–300 million (+16.5% YoY at midpoint); adjusted EBITDA margin 28–31%; adjusted EPS $0.40–0.44 (+68% YoY at midpoint). Forward P/E approximately 46×.

Cognex's vision systems are sold into logistics, automotive, EV, consumer electronics. Humanoid is potential upside but not a current line item; In-Sight + DataMan are gold-standard for industrial pick-and-place. Direct competition for humanoid head-mounted vision is more software-defined (NVIDIA Isaac Perceptor) but Cognex remains the incumbent for guided robotic applications.

Catalyst: Continued margin expansion; Q2 guide already raised; AI-vision product cycle.

Risk: Humanoid will use proprietary cameras (NVIDIA Holoscan, Intel RealSense legacy, custom) — not Cognex's traditional fixed-mount machine vision.

Verdict — Add (defensive): A quality industrial vision compounder having its best quarter sequence in years. Humanoid is upside, not the thesis. Size as a defensive industrial position with AI optionality.

HXGBY / HEXA-B.ST — Hexagon AB

Q1 2026: revenue EUR 964 million; +8% organic growth, flat reported. EBIT EUR 251 million, margin 26.1%. Gross margin 62.9%. The only Western large-cap with a named humanoid OEM line item — AEON humanoid built by Hexagon for industrial inspection / metrology / pick-and-place.

Customer disclosures: BMW Plant Leipzig — first European humanoid pilot (Feb 27, 2026 announcement; BMW chose AEON over Figure 03 for this site). Pilot completed; heading into production deployment. Schaeffler — long-term agreement for up to 1,000 AEON units over 7 years. Commercialization targeted by end of 2026.

Catalyst: AEON commercialization milestone; first BMW production deployment; additional auto OEM disclosures.

Risk: AEON is a single-tenant industrial robot story — not consumer or general-purpose. Revenue contribution will be small relative to Hexagon's EUR 4 billion+ base for several years.

Indian-resident accessibility: OTC HXGBY → IBKR. Vested / INDmoney coverage of Swedish OTC tickers is thin.

Verdict — Add (advanced): The only Western large-cap with a named humanoid OEM product line. Real and shipping but small contribution. IBKR-routed; not a Vested / INDmoney name.

KEYS, TDY, TRMB

Keysight Technologies (KEYS): Q2 FY26 revenue $1.72 billion (+30.5% YoY) record. Non-GAAP EPS $2.87. Orders above $2 billion. No humanoid revenue disclosed. Sells test & measurement to every semiconductor and SoC maker in this guide. Bull case is AI-test demand, not humanoid.

Teledyne Technologies (TDY): Q1 2026 revenue $1.56 billion (+7.6% YoY, +6.9% organic). Diversified defense / digital imaging / instrumentation. Humanoid is a tertiary thesis.

Trimble (TRMB): Q1 2026 revenue $939.9 million (+11.8% YoY). Positioning / GNSS / GIS for construction, agriculture, transport. Outdoor / construction robots are the adjacency, not bipedal humanoids.

Verdict — Skip (humanoid thesis): All three are real businesses but humanoid is not the right reason to own them. Buy KEYS for the AI-test thesis; buy TDY for instrumentation; buy TRMB for positioning. Not for humanoids.

Lidar — Ouster, Innoviz, Aeva: the explicit "do not chase for humanoid" call-out

Ouster (OUST): Q1 2026 revenue $49 million (+49% YoY). Industrial vertical is the strongest growth segment. Most humanoids use stereo / RGB-D cameras for primary perception, not lidar. Lidar is more for AMRs and outdoor bots. Real business; wrong story for humanoid.

Innoviz (INVZ): Q1 2026 revenue $7.1 million (NRE milestones pushed out). Pure automotive lidar (BMW L3, LOXO L4). Negative humanoid exposure right now.

Aeva (AEVA): Q1 2026 record revenue $6.3 million (+90% YoY). FMCW lidar primarily automotive plus Aeva CityOS smart infrastructure in Georgia. Not a humanoid play.

Luminar (LAZR): Filed Chapter 11 in December 2025. Volvo discontinued the relationship in April 2026. Subsidiary sales: Luminar Semiconductor to Quantum Computing for $110 million; lidar tech to Microvision for $33 million. Distressed.

Verdict (all four) — Avoid (humanoid thesis): Lidar is automotive, not humanoid. Don't buy OUST/INVZ/AEVA on a humanoid thesis. Avoid LAZR entirely.

ams OSRAM (AMS.SW)

Q1 2026 revenue EUR 796 million; adjusted EBITDA margin 16.5%. Restructuring story dominates — sold Specialty Lamps; selling non-optical mixed-signal business to Infineon. Optical sensors are humanoid-relevant for vision + ToF + proximity sensing but the corporate restructuring overwhelms the humanoid thesis. IBKR-only access.

Verdict — Skip: Restructuring story plus thin Indian retail access. Other names are cleaner.

Layer 3 top picks: Cognex for defensive industrial vision with cleanest fundamentals; Hexagon for the only large-cap with named humanoid OEM line.

Layer 4 — Power & battery: why humanoid is not a lithium thesis

The single biggest framework correction this guide can offer: a humanoid robot battery pack is approximately 2–3 kWh. An EV pack is 70–100 kWh. One million humanoids per year consumes roughly the lithium of 30,000 EVs. That is a marginal demand kicker, not a structural thesis.

Albemarle (ALB): Q1 2026 revenue $1.43 billion (+33% YoY); Energy Storage segment sales $891 million (+70%); adjusted EBITDA $664 million (+148% YoY). The lithium recovery is real but the humanoid attribution is small. FY26 guidance ranges from $0.9 billion to $4.4 billion adjusted EBITDA depending on lithium price scenario.

Lithium Americas (LAC): Pre-production; Thacker Pass mechanical completion targeted late 2027; cash $1.21 billion; FY26 capex guide $1.3–1.6 billion. Not a humanoid play.

Sigma Lithium (SGML): Q1 2026 revenue $42.3 million; net income $11.1 million; gross margin 61%, EBITDA margin 39%. Production 23.6kt 5% lithium concentrate. Not a humanoid play.

SQM: Q1 2026 revenue $1.76 billion (+69.8% YoY); net income $364.7 million (+165.2%); EPS $1.28. Lithium revenue $1.19 billion (+135.7%); volumes ~69kt LCE at full capacity; ASP approximately $18/kg in Q1 vs $10/kg in Q4 2025. Not a humanoid play.

LG Energy Solution (KRX 373220): Q1 2026 revenue KRW 6.6 trillion (+1.2% QoQ); operating loss KRW 207.8 billion. Total backlog above 440 GWh by April. ESS now mid-20% of revenue. Korean-listing; IBKR only. Not a humanoid play.

Verdict (all five) — Skip (humanoid thesis): Lithium is an EV / ESS thesis. If you want lithium exposure, take it for those reasons. Do not buy any of these names "for humanoids" — the math doesn't support it.

Layer 5 — The OEMs: Tesla and Boston Dynamics are the only public-market humanoid OEMs that matter

Almost every credible humanoid OEM in 2026 is private. Figure ($39 billion post-money valuation as of September 2025), 1X Technologies (raising to $10 billion+), Apptronik ($5 billion as of February 2026), Agility Robotics, Sanctuary AI, Unitree (Chinese), Sharpa — none of these are accessible to Indian retail investors. The two public-market OEM exposures are Tesla (Optimus) and Hyundai (through Boston Dynamics).

TSLA — Tesla (Optimus)

Q1 2026: total revenue $22.39 billion (+16% YoY, missed $22.64B consensus); non-GAAP EPS $0.41 (beat $0.34 by ~21%); GAAP gross margin 21.1% (+478 bps YoY). FY26 capex raised to above $25 billion from the prior $20 billion. Forward P/E approximately 210–215× — extreme even by Tesla standards.

Optimus production status from the Q1 2026 earnings call:

Several hundred units produced in Q1 2026
Characterized by Musk as "primarily for learning, not productive tasks — still very much in R&D phase"
Fremont conversion: Last Model S/X off the line early May 2026; Optimus production line targeted to begin late July–August 2026. Designed first-gen capacity: 1 million units/year at Fremont.
Giga Texas: Site preparation underway; second-gen Optimus line targeting 10 million units/year; production expected summer 2027.
Musk: initial Fremont output "quite slow," "literally impossible to predict" — 10,000 unique parts on a new line.
Optimus V3 reveal pushed later in 2026 (multiple delays).

Production target disclosed: 100,000–300,000 units in 2026 per Musk. Realistic shipped count for 2026: low thousands by year-end. Deutsche Bank consensus puts total industry humanoid units at 200,000 shipped by 2035 — i.e., Tesla's stated single-year 2026 target equals Deutsche Bank's industry consensus for the entire decade through 2035. Pricing target: $20,000 retail at mass production (confirmed at the 2025 shareholder meeting); public availability targeted late 2027 per Musk at Davos 2026.

Bank of America (March 2026 resumed coverage): Values Optimus division at above $30 billion, approximately 2% of TSLA market cap. Price target $460–$471.

Catalyst: Optimus V3 reveal; Fremont production start (August); first external customer if any; Q3/Q4 unit count disclosure on earnings calls.

Risk (humanoid-specific): Optimus is priced into TSLA at above $30 billion by BofA already. If V3 disappoints or Fremont ramp slips into 2027, multiple compression is severe given 210× forward P/E. Optimus-as-bull-thesis is fully priced.

Verdict — Hold (existing) / Watch (new): If you already own TSLA for autos + autonomy + Optimus, hold. New positions are paying a 210× forward P/E for an Optimus catalyst that has slipped multiple times. Wait for Fremont production proof in Q4 2026 before initiating.

HYMTF / 005380.KS — Hyundai Motor (owns Boston Dynamics)

Q1 2026: revenue KRW 45.94 trillion (record, +3.4% YoY). Operating profit KRW 2.51 trillion (−30.8% YoY on tariffs). Global vehicle sales 976,000 (−2.5%). Electrified vehicles 24.9% of mix.

Boston Dynamics Atlas Electric program — the headline humanoid disclosure:

Production-ready Atlas unveiled at CES 2026 (Jan 5, 2026)
Entire 2026 production run already committed to two customers: Hyundai Motor Group + Google DeepMind
Deployment sites: Hyundai Robotics Metaplant Application Center (RMAC), Bryan County, Georgia ($5.5 billion greenfield EV assembly plant designed for robotic flexibility)
Hyundai-Boston Dynamics announced $26 billion US investment including a new robotics factory designed for 30,000 robots/year capacity
Fully electric (vs prior hydraulic Atlas) — every joint electrically actuated

Catalyst: Atlas commercial deployment milestones; Georgia robotics factory groundbreaking; first external Atlas customer if any.

Risk: Hyundai is fundamentally an auto company. Atlas / Boston Dynamics is single-digit percentage upside on the equity even in a bullish scenario. Tariff headwinds dominate near-term.

Indian-resident accessibility: OTC HYMTF → IBKR + some platforms. KRX 005380 → IBKR only. Liquidity in HYMTF is thin.

Verdict — Add (advanced, IBKR): The only large-cap public route to Boston Dynamics. Atlas is the only humanoid program with 100% of 2026 production pre-committed. Cheap Korean auto multiple gives downside cushion. Indian access is operationally messy via OTC ADR.

Xiaomi (HKEX 1810) and BYD (HKEX 1211) — the Chinese OEMs

Xiaomi: CyberOne / CyberOne V2 in internal R&D + factory pilot. CyberOne deployed in Xiaomi EV die-casting workshop with 90.2% success rate, ~76-second cycle time, 3-hour autonomous operation runs. Internal pilot only — not sold commercially.

BYD: Launched 7th-generation humanoid robots for global 4S dealership showrooms (Q2 2026); internal codename Yao-Shun-Yu project. Production target: 20,000 humanoid units in 2026. UBTech Walker S1 used at BYD factories.

Both are HKEX-listed. Indian access: IBKR primary; OTC BYDDY available on most platforms.

Verdict — Watch: Real corporate strategy but humanoid is a far-too-small share of equity to move the stock in 2026. Buy these for EV / smartphones / Chinese consumer thesis, not for humanoids.

The names you should not own for humanoids in this layer

Honda (HMC): ASIMO program officially cancelled in 2022. "ASIMO" is now rebranded as ASIMO OS — a vehicle operating system for Honda 0 Series EVs starting 2026. Honda no longer has an active humanoid program. Not a humanoid play.
Toyota (TM): Punyo is a Toyota Research Institute research platform — soft-bodied torso-up humanoid for whole-body manipulation. Pure research; no production timeline. Toyota / TRI partnered with Boston Dynamics on Atlas in 2024 but that is a research collaboration. Not a humanoid play.

Layer 5 top pick: HYMTF for the cleanest "we already make the humanoid" exposure at a depressed auto multiple. TSLA only if the Optimus catalyst is the marginal thesis you're chasing and you accept the 210× forward P/E.

Layer 6 — Adjacent industrial robotics, simulation, and structural changes you should know

This layer is mostly the "looks like humanoid but isn't" pile. A few real industrial automation incumbents, and several names that have undergone structural changes that remove them from a humanoid thesis entirely.

SYM — Symbotic

Q2 FY26 (March quarter) revenue $676 million (+23% YoY); net income $9 million; adjusted EBITDA $78 million (above 2× YoY); backlog $22.7 billion. Walmart anchor customer ($520 million stake; 40 BreakPack bots per site order). Forward P/E approximately 111–128× (modestly overvalued per GuruFocus GF Value of $45 vs price ~$57).

Symbotic is warehouse automation (case picking + palletization), not humanoid. The bot fleet is wheeled and fixed, not bipedal. Walmart concentration is single-customer risk. Recent acquisition: Fox Robotics (dock automation).

Verdict — Skip (humanoid thesis): Real warehouse robotics business but not a humanoid pure-play. If you want warehouse automation exposure, look at SYM on its own merits — not as humanoid optionality.

ROK, EMR, HON — the industrial automation incumbents

Rockwell Automation (ROK): Q2 FY26 revenue $2.24 billion (+12% reported, +9% organic). Adjusted EPS $3.30 (+32% YoY, beat $2.88 consensus). Pre-tax margin 19.7% vs 14.9% prior year. End markets warehouse automation, data center, semiconductor, energy all improving. Humanoid is upstream (sells motion control + PLCs into humanoid OEMs and into factories deploying humanoids).

Emerson Electric (EMR): Q2 FY26 revenue $4.56 billion (+3% reported, +0.5% organic). Process automation (LNG, life sciences, chemical). Not a humanoid play.

Honeywell (HON): Q1 2026 revenue $9.1 billion (+2% organic). Industrial Automation $1.4 billion (+1% organic). Strategic context: aerospace spin-off targeted Q3 2026; sale of Productivity Solutions + Warehouse Workflow Solutions also closing 2026. Honeywell is actively reducing warehouse / robotics exposure. Not a humanoid play going forward.

Verdict — Skip (humanoid thesis): Real industrial businesses but humanoid is not the right reason to own them. ROK is the closest to humanoid-adjacent if you want indirect exposure.

ABB — structurally different post-SoftBank divestiture

Q1 2026: revenue $8.7 billion (+18% reported, +11% comparable). Operational EBITA $2 billion+ (+37%), margin 23.5%. Book-to-bill 1.29; backlog $27.5 billion record. Net income $1.32 billion.

Critical structural news: ABB has signed an agreement to divest its Robotics division to SoftBank Group for an enterprise value of $5.375 billion (announced Q1 2026, replacing the earlier spin-off plan). Expected to close mid-to-late 2026. Post-divestiture, ABB's go-forward business will not include industrial robotics.

Verdict — Avoid (humanoid thesis): ABB is no longer a humanoid play after the SoftBank divestiture closes. The acquirer (SoftBank) is the one to watch; ABB will be electrification and process automation only.

FANUY / 6954.T, Kawasaki, Yaskawa — the Japanese industrial robot incumbents

Fanuc (TSE 6954 / ADR FANUY): FY March 2026 revenue ¥857.83 billion (+7.62% YoY). Robot Division ¥269.241 billion (+11.1% YoY) driven by China EV demand. NVIDIA Physical AI partnership announced 2025/2026 — Fanuc as global robotics ecosystem partner.

Kawasaki Heavy (TSE 7012 / OTC KWHIY): FY March 2026 revenue ¥2.31 trillion (+8.54% YoY). Precision Machinery & Robot segment revenue ¥179.53 billion (+16% YoY); profit forecast nearly doubling next year (¥7.0B → ¥14.3B). Conglomerate dilution — aerospace + hydrogen + shipbuilding dominate.

Yaskawa Electric (TSE 6506 / OTC YASKY): FY March 2026 revenue ¥542.1 billion; net income ¥35.2 billion (net margin 6.5%, down from 11% prior FY). Robotics segment ¥247.01 billion. FY27 guide: net sales ¥580B; OI ¥60B.

Verdict (all three) — Watch: Industrial robot incumbents, not humanoid OEMs. Fanuc is the cleanest pick if you want indirect humanoid exposure via the NVIDIA Physical AI partnership. All IBKR-routed for Indian residents.

The structural changes you need to know

Synopsys completed the $35 billion acquisition of Ansys in 2026. Ansys is now Synopsys's Design Automation segment. Synopsys Q1 2026 revenue $2.41 billion (+65.4% YoY); Ansys contributed $886 million (~37% of total). FY26 guide revenue $9.61 billion midpoint. Multibody dynamics and FEA simulation is humanoid-relevant but is a small portion of the Ansys book inside the larger SNPS business.
iRobot (IRBT) emerged from Chapter 11 January 23, 2026. Picea acquired 100% equity. No longer publicly listed in the traditional sense. Avoid.
Luminar (LAZR) — Chapter 11 December 2025 (covered in Layer 3). Avoid.

Cadence (CDNS), Bentley (BSY), Autodesk (ADSK)

Real businesses but humanoid is a thin tertiary thesis. Cadence's recent Hexagon Design & Engineering acquisition explicitly positions for physical AI; Q1 2026 revenue $1.47 billion (+18.7% YoY); FY26 guide raised to +17% growth. But the bigger driver is AI chip design demand.

Verdict — Hold (CDNS only, semi-thesis): Cadence works on its own merits as covered in the semiconductor guide. Humanoid is a free option.

Intuitive Surgical (ISRG)

Q1 2026 revenue $2.77 billion (+23% YoY). Robotic surgery, not humanoid. Da Vinci procedures +16%; Ion procedures +39%. Not a humanoid play.

Verdict — Skip (humanoid thesis): Buy ISRG for surgical robotics if you want it. Not for humanoid.

Three model portfolios

Sketches, not advice. Position weights are within the humanoid allocation — not your full portfolio. Sizing thematic humanoid exposure inside your overall asset mix is the question above this one. As covered in the semi guide, a widely cited institutional heuristic caps single-sector thematic exposure at 15% of an equity portfolio, single names at 5%, combined sector exposure at 20%. Humanoid is more speculative than diversified semi — most investors should size this layer smaller than they would semis.

A real constraint: of the names I'd actually pick for a humanoid portfolio, several are IBKR-only (Schaeffler, Harmonic Drive, Hexagon, Boston Dynamics via Hyundai). If you're on Vested / INDmoney / Rovia only, you cannot replicate the cleanest humanoid portfolios without opening IBKR. That's a real friction.

Defensive: large-cap with non-humanoid cushion

For investors who want humanoid optionality but cannot tolerate single-thesis drawdowns. Most positions are owned for other reasons anyway — humanoid is the free option.

Layer	Name	Weight	Indian-access
Compute	NVDA	25%	All platforms
Compute	QCOM	15%	All platforms
OEM	TSLA	12%	All platforms
OEM	HYMTF	8%	IBKR primary
Sensors	CGNX	10%	All platforms
Adjacent	FANUY	8%	IBKR best
Adjacent	ROK	7%	All platforms
Cash buffer	—	15%	—

Logic: 40% in compute (NVDA + QCOM) — both work even if humanoid TAM disappoints. 20% in OEMs at half the conviction weight given Tesla's 210× P/E and Hyundai's tariff headwinds. 18% in sensors + adjacent industrial automation. Cash buffer for the inevitable humanoid-cycle disappointment.

Balanced: adds the actuator layer and direct humanoid exposure

For investors with IBKR access who want real humanoid pure-play exposure alongside core compute.

Layer	Name	Weight	Indian-access
Compute	NVDA	18%	All platforms
Compute	QCOM	10%	All platforms
Compute	AMBA	4%	All platforms
Actuators	SCFLF (Schaeffler)	10%	IBKR
Actuators	HSCDF (Harmonic Drive)	8%	IBKR
Actuators	6479.T (Minebea Mitsumi)	4%	IBKR
Sensors	CGNX	8%	All platforms
Sensors	HXGBY (Hexagon)	7%	IBKR
OEM	TSLA	10%	All platforms
OEM	HYMTF	6%	IBKR primary
Adjacent	FANUY	5%	IBKR best
Cash buffer	—	10%	—

Logic: 32% compute, 22% actuators (the BOM math says this is the highest-direct-exposure layer), 15% sensors, 16% OEMs. IBKR routing required for 35% of the portfolio.

Aggressive: leverage to the actuator BOM math and humanoid catalysts

For investors with IBKR access who want maximum humanoid exposure and can tolerate 50%+ drawdowns. The actuator layer takes the largest weight because actuators are 56% of BOM — disproportionate to current revenue but disproportionate to forward humanoid revenue too.

Layer	Name	Weight	Indian-access
Actuators	SCFLF (Schaeffler)	18%	IBKR
Actuators	HSCDF (Harmonic Drive)	15%	IBKR
Compute	NVDA	14%	All platforms
Compute	QCOM	9%	All platforms
Sensors	HXGBY (Hexagon)	8%	IBKR
OEM	TSLA	8%	All platforms
OEM	HYMTF	7%	IBKR primary
Sensors	CGNX	6%	All platforms
Specialty	AMBA	4%	All platforms
Adjacent	FANUY	4%	IBKR best
Cash buffer	—	7%	—

Logic: 33% actuator layer (Schaeffler + Harmonic Drive) as the BOM-percentage-tracking conviction allocation. 23% compute. 14% sensors. 15% OEMs. Expected drawdown in a humanoid TAM disappointment scenario: 50–65%.

What not to chase

The categories that look like humanoid plays in social media coverage but carry more risk than the multiples compensate for:

Lidar names (OUST, INVZ, AEVA) — automotive plays mis-tagged as humanoid. Most humanoids use stereo / RGB-D cameras, not lidar.
Luminar (LAZR) — Chapter 11. Avoid entirely.
iRobot (IRBT) — Chapter 11, no longer traditional listing. Avoid.
Honda (HMC) — ASIMO cancelled in 2022. Not a humanoid play.
Toyota (TM) — Punyo is research only. Not a humanoid play.
Lithium names (ALB, LAC, SGML, SQM) — humanoid is a marginal demand kicker on EV / ESS. Don't buy these for humanoids.
ABB — post-SoftBank divestiture, no longer a robotics company.
Honeywell (HON) — actively divesting warehouse + robotics business.
Synopsys (SNPS) post-Ansys — humanoid is a thin slice of a larger EDA business.
Marvell (MRVL), Astera Labs (ALAB) — AI data center compute, not humanoid.
Intuitive Surgical (ISRG) — surgical robotics, separate thesis.

The risk scenarios you should price in

Five frameworks that materially change the picture.

Capability vs timeline expectations (the historical track record). ASIMO ran 2000–2022 and never commercialized. Atlas hydraulic ran 2013–2024 before being replaced by Atlas Electric. Tesla Optimus was announced August 2021; five years later Q1 2026 production was several hundred units for R&D. Every prior humanoid program has been late vs initial promise by 3–7× on timeline. Goldman's 1.4 million units by 2035 implies a 70% CAGR; if delivery slips even two years, the cumulative number compresses dramatically.

Substitution risk (the boring competitors). AMRs (Symbotic, Locus, Geek+ — wheeled, no balance complexity) outcompete bipedal humanoids for above 80% of warehouse use cases. Cobots (Fanuc, Yaskawa, ABB → SoftBank, Universal Robots) are mature and cheaper. Toyota's Punyo soft-bodied approach has different ROI envelope. If a task can be done by a non-humanoid form factor, it will be. The defensible humanoid TAM is narrower than the $5 trillion Morgan Stanley headline.

TAM realization risk. Morgan Stanley's $5 trillion by 2050 assumes humanoids substitute non-trivial fractions of US ($28/hr average factory worker) and global manufacturing + service labor at $20–40k unit cost with operating uptime matching or exceeding human (the unsolved problem as of 2026). The cited $5 trillion includes services + supply chain + repair + AI/software — i.e., it is total addressable economy, not OEM market revenue. Specific BOM-revenue capture by listed-equity OEMs is a far smaller number than the headline.

China-vs-US cost dynamic. Unitree G1 at $13,500–$16,000 retail in 2026 sets the cost ceiling. Tesla's $20,000 target is now competitive parity, not cost leadership. Tesla Optimus BOM is $46,000 with Chinese supply chain or $131,000 without per analyst teardowns — i.e., Tesla is structurally dependent on Chinese components for cost competitiveness. US humanoid OEMs face structural cost disadvantage in commodity components; differentiation must come from AI stack (NVIDIA-anchored) or vertical integration (Tesla model).

Regulatory / safety framework. ISO 13482:2014 is the existing service robotics safety standard but lacks enforcement. Real regulation likely from OSHA (industrial deployments) and EU AI Act (humanoid + autonomy classification). Mercedes-Benz / BMW pilots are running under industrial safety standards (existing factory protocols) — household deployment will trigger a different regulatory framework entirely. Regulatory cost is undefined.

How an Indian-resident investor actually executes this

The IBKR question. Of the cleanest humanoid pure-plays — Schaeffler (XETRA), Harmonic Drive (TSE), Hexagon (Swedish OTC), Minebea Mitsumi (TSE), Fanuc (TSE direct or OTC ADR), Boston Dynamics via Hyundai (Korean direct or OTC ADR) — most route through IBKR. Vested, INDmoney, and Rovia all support US-listed names (NVDA, QCOM, TSLA, CGNX, AMBA, SYM, CDNS) but coverage of OTC ADRs is variable and direct XETRA / TSE / KRX / HKEX listings are typically not supported. For real humanoid exposure beyond the US-listed names, you likely need to open IBKR. The structural cost comparison between platforms is in Vested vs INDmoney vs IBKR vs Rovia.

Tax on gains. Foreign equity LTCG = 12.5% under Section 112 with a 24-month holding period. STCG = slab. Indexation removed for foreign equity post-Budget 2024. Applies equally to US listings and to direct foreign listings (TSE, XETRA, KRX, HKEX).

Schedule FA. Any of these names held at any point during the calendar year (January–December) triggers a Schedule FA disclosure obligation. Penalty for non-disclosure under the Black Money Act 2015: 30% tax + 3× penalty + 3–10 year prosecution risk. See the Schedule FA disclosure guide.

Dividend withholding by listing geography. US listings 25% under India-US DTAA with valid W-8BEN (credit via Form 67 / Form 44 from TY2026-27). Japan 15% under India-Japan DTAA. Korea 15% under India-Korea DTAA (capital gains generally not taxed in Korea for non-residents per DTAA). Hong Kong: no dividend withholding; no capital gains tax. Switzerland 35% (recoverable to 10% under DTAA via refund process — operationally painful for retail). Germany 26.375% (recoverable to 10% under DTAA).

Practical implication for Indian investors. OTC ADR route is operationally simpler than direct foreign listings (one withholding rate, no refund process), despite slightly lower liquidity. Only choose direct foreign listing where ADR is not available (Schaeffler's OTC SCFLF is illiquid; XETRA SHA.DE is the cleaner route via IBKR) or where the ADR trades at a meaningful discount.

Currency exposure. Long USD / EUR / JPY / KRW / HKD against the rupee depending on the listing. The 24-month holding period for LTCG locks in that currency exposure. Build into position sizing.

The closing

The honest read on June 2026:

Zero public companies separately disclose humanoid revenue. Schaeffler's "first series production Q2 2026" is the closest thing to a forward humanoid revenue commitment.
Actuators are 56% of every humanoid BOM. Buying Schaeffler and Harmonic Drive is the most direct supply-chain bet. Both require IBKR.
NVIDIA, Qualcomm, and Tesla are owned for other reasons — humanoid is the free option. Don't size them as if humanoid revenue is real.
Boston Dynamics inside Hyundai is the only large-cap "we make the humanoid" play outside Tesla. Atlas Electric pre-committed all 2026 production to Hyundai + Google DeepMind. Indian access via IBKR.
Most names tagged as humanoid plays in coverage aren't. Lidar is automotive. Lithium is EV. ABB sold its robotics. Honda cancelled ASIMO. Honeywell is divesting warehouse robotics.

The category is real. The TAM is huge on aggressive assumptions, narrower on honest ones. The companies most likely to actually capture humanoid revenue per unit shipped are the actuator suppliers, not the AI compute giants. The Indian retail investor who wants real exposure needs IBKR.

After the rally is the worst time to chase. It is the best time to buy the businesses that will ship the actual parts of every humanoid produced, regardless of which OEM wins.

The forums see the humanoid videos. The order books are in the strain wave gears.