ARK Autonomous Technology & Robotics ETF
ARK Invest — Cathie Wood
Active fund (≥80% of assets) in autonomous transportation, robotics & automation, 3D printing, energy storage and space; ~41 holdings. Expense ratio 0.75%.
Top holdings
Top position — autonomy/robotaxi + humanoid-robot thesis
AI/robotics compute exposure
Automated test + industrial robotics (Universal Robots/MiR)
Small-launch + space-systems name
Autonomous/unmanned defense systems
Performance
July 2026: 12.97% (Yahoo adj close, $132.22 on 30 Jun -> $115.07 on 31 Jul); TradingView -12.54% — July's AI-hardware rout hit ARKQ harder than its theme labels suggest: -12.97%, erasing the entire 2026 gain (+15.3% YTD at 30 Jun -> +0.36% at 31 Jul). The worst movers were the distinctively-ARK positions, not just semis: SPCX -36.6%, RKLB -36.1%, TSLA -26.0% (Q2 miss on AI/robotics spend), TER -24.0%, AMD -18.0%, TSM -15.4%. Offsets: AMZN +14.0%, PLTR +5.5%, GOOG +0.9%, NVDA +0.3%.
Recent moves
Leaned further into defense-autonomy and space (Kratos, Rocket Lab) while keeping Tesla and AMD as core compute/robotics anchors.
Our take
The cleanest ARK vehicle for the physical-AI thesis (robotics, autonomy, defense-tech) and a notably strong 1-year performer — more thematically coherent than ARKK and a credible way to own the 'AI leaves the data center' trade.
Top ten as of 20 Jul 2026: Tesla 10.77%, AMD 6.56%, SpaceX 5.91%, Teradyne 5.83%, Kratos 5.49%, Alphabet 5.05%, Palantir 3.88%, Deere 3.65%, Rocket Lab 3.55%, TSM 3.39%.
The barbell is real but smaller than either ARK's critics or its marketing suggest. The cheaply-replicable large-cap sleeve — Alphabet, TSM, AMD, Deere — is 18.65% of assets; the ARK-flavored defense-and-space cohort — SpaceX, Kratos, Rocket Lab, Palantir — is 18.83%. Nearly the same size, and together only ~37% of the fund. Tesla at 10.77% belongs to neither: it is the house view, the single-name expression of the autonomy thesis ARK has run since 2014, and the largest position by a wide margin. Teradyne at 5.83% is the most interesting holding and the one the barbell story usually drops. Automated-test equipment is neither mega-cap ballast nor a headline theme trade — it is where discretionary selection would actually be doing work. So: a house-view Tesla bet, a defense/space sleeve, an index-like sleeve, and a few genuine picks — priced at 0.75% across all four.
- 2026-07At 30 Jul 2026: TSLA 9.78%, SPCX 6.41%, TER 6.03%, AMD 5.81%, KTOS 5.72%, GOOG 5.07%, PLTR 3.95%, DE 3.59%, TSM 3.50%, LHX 3.40%; top 10 53.25%. AMZN 3.31% and NVDA 3.22% now widen the replicable sleeve.
Thesis
ARKQ's premise is that AI's next value capture happens outside the data center — in machines that move: autonomous vehicles, industrial robots, drones and defense autonomy, reusable launch. It expresses this actively (41 holdings, no index) on the bet that a discretionary manager owns the winners before a robotics benchmark admits them. It is the most thematically coherent ARK vehicle, but coherence and correctness are different claims — and the 2026 record tests the second one.
Actively managed, concentrated in autonomous transport, robotics/automation, 3D printing, energy storage and space. 41 positions, top 10 at 54.10% of assets. No leverage, no derivatives. Turnover is conviction-driven: ARK adds into drawdowns in names it already owns rather than trimming to a target weight. Expense ratio 0.75%, AUM $1.92B. (Figures as of 20 Jul 2026, stockanalysis.com.)
Assessment
- Genuinely coherent theme: every top-ten holding maps to a machine that moves — launch, drones, robots, autonomy — or the silicon driving them.
- Owns the defense-autonomy cohort (Kratos 5.49%, Rocket Lab 3.55%, Palantir 3.88%) that passive robotics indices under-weight or exclude on liquidity screens.
- Holds SPCX at 5.91%, a launch exposure most robotics ETFs have at zero weight.
- 41 holdings with top 10 at 54.10% is concentrated enough to matter but far from a five-stock bet; a single-name blowup is survivable.
- Supplied context figures (~$1.5B AUM, ~+45% 1yr) do not survive checking: stockanalysis.com shows $1.92B AUM and 16.98% 1-year total return as of 20 Jul 2026.
- A ~18.65% sleeve of Alphabet, TSM, AMD and Deere is cheaply replicable large-cap growth, yet the 0.75% fee is charged on the whole portfolio, not just the discretionary part.
- Tesla at 10.77% is a decade-old house view whose autonomy monetization remains the fund's central unresolved claim — the largest position is also the least diversifiable one.
- ARK's conviction-averaging means the fund adds into losers — a feature if the thesis is right, a compounding hole if not, with no mechanical stop.
- 2026 has been roughly flat: the trailing-year gain was earned before January, so the theme cohort is not currently re-rating the way the narrative implies.
- 2026-07The defense-and-space cohort was July's drag rather than its differentiator: SPCX -36.6%, RKLB -36.1% — Rocket Lab fell to #13 at 3.17% and L3Harris entered the top 10 at 3.40%. Only Kratos and Palantir held up.
Record
Since-inception average annual return is 17.05% (inception 30 Sep 2014) and the trailing year 16.98%, both as of 20 Jul 2026. But 2026 itself has been roughly flat — +1.29% YTD as of 17 Jul (totalrealreturns.com). That inverts the obvious story: essentially all the trailing-year gain predates January, and the defense/space/robotics cohort has not driven a 2026 re-rating. The peer split is the useful attribution. Over the same trailing year, BOTZ (large-cap robotics index, 0.68%) returned 3.79% and ARTY (AI-select index, 0.47%) returned 55.50%. ARKQ's 16.98% sits between — well ahead of pure robotics, far behind pure AI. What paid was AI exposure, not robotics, and ARKQ captured a fraction of it via AMD, TSM and Alphabet. The distinctively-ARK sleeve is not what produced the number. The 17.05% since-inception figure also spans a 2020 melt-up and a 2022 drawdown; a dollar-weighted investor arriving after the run earned far less.
Risks & fit
- Concentration: top 10 at 54.10%, Tesla alone at 10.77% — fund returns are hostage to a handful of high-volatility positions.
- Kratos and Palantir carry appropriations and procurement-cycle risk that is political, not technological — a US budget shift hits several holdings at once.
- SPCX at 5.91% is a recently-listed position with no long public trading history to anchor a valuation range.
- Key-person risk: the strategy is Cathie Wood's discretionary framework, with no articulated succession for the research view.
- Mid-cap space/defense holdings mean fund flows move the underlying — outflows force selling into thin books.
The thesis breaks if returns keep decomposing into holdings you could own cheaply: if over 12-24 months a low-cost blend of a robotics index and an AI/semis index matches or beats ARKQ net of 0.75%, the discretionary layer adds nothing. The trailing year already points that way — 16.98% sits between BOTZ's 3.79% and ARTY's 55.50%, consistent with partial AI beta rather than selection skill. Second falsifier: if humanoid and industrial robotics deployment slips another cycle while Tesla autonomy revenue stays deferred, the premise loses its clock and the fund is a high-beta defense proxy in a robotics label.
An allocator who already holds core equity, wants deliberate tilt toward defense autonomy, launch and physical robotics, and can hold through deep drawdowns without resizing — a satellite-sized, multi-year exposure. It does not suit anyone needing it to behave like diversified tech, anyone whose reason for owning it is the AI trade specifically (a cheaper AI-select fund gave more of that over the last year), or anyone who would be forced to sell in a growth-factor drawdown.
0.75% versus 0.68% for BOTZ and 0.47% for ARTY (as of 20 Jul 2026). The premium buys discretionary selection, but is levied on the whole portfolio — including the ~18.65% Alphabet/TSM/AMD/Deere sleeve where active judgment does least work.
- 2026-07The asymmetry is worse than partial AI beta: in July ARKQ fell -12.97%, more than pure-AI ARTY (-12.5%) and nearly double robotics BOTZ (-7.2%), after capturing roughly a third of ARTY's 1-year gain.