§ 01 — The Financial Position
$1.3 billion in Tesla stock
ARK's 13F-HR filings with the SEC show Tesla as a top-three holding across the firm's liquid funds
in every quarter since 2020. ARK earned management and performance fees on this position for five years.
When ARK publishes research describing Tesla's humanoid robot as a $26 trillion market opportunity,
that research is produced by a firm with more than a billion dollars riding on whether the thesis is correct.
ARK ARKK top holdings by approximate value — Q4 2025 13F (filed Feb 11, 2026)
Tesla position as filed: 2,914,492 shares · $1.31B
per Q4 2025 13F-HR filed February 11, 2026. ARK trimmed the position by 18.8% during Q4 2025
(from 3.59M shares). Tesla remains the single largest liquid equity holding in ARKK.
ARK ARKK AUM — quarterly peak to Q4 2025 ($B)
−72% from peak
ARK's AUM peaked at $53.74B in Q2 2021. By Q4 2025 it had fallen to approximately $15.07B,
a 72% decline. Tesla's share price decline accounts for a large portion of this erosion.
ARK's fee income depends on AUM. AUM depends heavily on Tesla.
ARK Tesla price targets — published vs. actual price
Every target missed
ARK published a $4,000 Tesla target for 2024 (missed by approximately 89%). The current
$2,617 target for 2029 requires a 63.5% annual compound growth rate from the
Q4 2025 price. No prior ARK Tesla target has been met on its stated timeline.
The structural incentive in plain terms: ARK charges a management fee, typically
0.75% to 0.87% annually, on AUM. On a $1.31 billion Tesla position (Q4 2025 13F, ~$451/share),
that generates approximately $9.8 million to $11.4 million in annual fee income from that single
holding alone. At the May 2026 price of $396.64, the position is worth approximately $1.15 billion,
generating approximately $8.6 million to $10 million annually. If ARK's Tesla thesis
fails and the stock declines, fee income falls. If ARK's bullish research attracts retail investors
who buy ARKK, AUM rises and fee income rises. ARK's own fund disclosures state it has "an incentive
to show the best performing results." This is not unusual for fund managers. The question is whether
that incentive shapes the framing of research about a company that represents the fund's largest bet.
§ 02 — The Research Narrative
Tesla leads. Competitors footnoted.
ARK's published humanoid robotics research frames the market as one where Tesla has a decisive
structural advantage. The language is specific: full stack AI capabilities, autonomous driving
expertise, and manufacturing scale. Figure, Apptronik, and 1X, the three companies ARK holds
venture stakes in, appear briefly, if at all, in ARK's public narrative about who will win the market.
What ARK says publicly about Optimus
ARK's Big Ideas 2025/2026 and published research articles describe humanoid robotics as a
$26 trillion global revenue opportunity with Tesla positioned as the leader. The specific
advantages cited: FSD AI stack applicable to robot training, Dojo supercomputer for
inference, vertical manufacturing integration, and Musk's "track record of execution
in complex hardware." ARK's 2029 Tesla price target allocates roughly 10% of value to
Optimus and the remaining 90% to robotaxi. The humanoid thesis supports valuation
indirectly by amplifying the AI narrative.
What ARK does not say publicly
ARK's venture fund holds equity positions in Figure AI (participated in the $675M
Series B, February 2024), Apptronik (participated in $403M Series A led by Google,
March 2025, and $520M extension, February 2026), Agility Robotics, and 1X Technologies.
After Figure's September 2025 Series C raised $1 billion at a $39 billion valuation,
ARK's Figure stake reportedly appreciated approximately 15x. ARK is a significant
financial beneficiary of Figure's success. Its public research does not reflect this.
The disclosure gap: ARK Venture Fund is a separate vehicle from ARKK and
ARKQ. The venture positions in Figure, Apptronik, and 1X are not required to appear in
quarterly 13F-HR filings, which only capture publicly traded securities. Retail investors
who follow ARK's public research on humanoid robotics have no visibility into ARK's
financial stake in Tesla's named competitors. ARK's public website and published articles
do not disclose these venture positions when discussing the humanoid robotics market.
The firm publishes a bullish Tesla narrative while holding
undisclosed financial stakes in the companies its research omits.
§ 03 — Deployment Reality
one robot is in production. the other is "for learning."
In May 2026, one company has robots on a factory floor generating commercial output. The other
has robots in its own facilities that its CEO describes as operating "primarily for learning."
ARK's research does not make this distinction.
Internal units deployed
Several hundred
CEO's own description
"primarily for learning"
2025 production target
5,000 units
Actual 2025 production
~hundreds (90%+ miss)
Commercial sales ETA
Late 2026 at earliest
Verified external ROI
None disclosed
Customer deployment data
None
External deployments
BMW Spartanburg: 40 units
Vehicles contributed to
30,000+ BMW X3 vehicles
Components moved
90,000+ (mm-accurate)
Operating hours logged
1,250+
Operating cost est.
~$25/robot-operating-hour
Deployment ramp time
12 mo. (1st), 30 days (2nd)
Verified customer ROI
Documented at BMW
The 30-day deployment curve
Figure's most significant unreported competitive advantage is its deployment learning rate.
Its first BMW use case required 12 months from initial engagement to live production.
The second customer took 30 days. That compression, from 12 months to 30 days, is a direct
consequence of real world deployment data. Each factory teaches the robot system faster than
simulated training can. Companies with robots in live production accumulate this data.
Companies with robots in internal R&D do not.
Tesla's Optimus fleet, operating internally at Fremont and Giga Texas, may be accumulating
training data. But Tesla has not disclosed what tasks Optimus performs, at what success rate,
or what production yield those robots achieve. ARK's research does not distinguish between
a robot in an R&D lab and a robot generating verifiable commercial output.
Optimus milestone record
At Tesla AI Day 2022, Musk stated that Optimus would be ready for production within a year.
By end of 2023, Tesla had demonstrated prototype units in controlled settings but had
not produced robots for external commercial sale. ARK's 2022 and 2023 research cited
the 2023 timeline as a near term catalyst. The timeline did not materialize.
Tesla projected deploying thousands of Optimus units in its own factories by end of 2023.
The actual count was in the tens of units, all in prototype or evaluation status.
No production-grade Optimus unit was deployed for repeatable commercial tasks in 2023.
Tesla's 2025 Optimus production target was 5,000 units. On the Q4 2025 earnings call,
Musk confirmed that "several hundred" were deployed, "primarily for learning, not
productive tasks." This is a miss of more than 90% on the stated annual target.
ARK's 2029 valuation model has Optimus contributing meaningfully to Tesla's revenue
starting in the 2026 to 2028 timeframe. The 2025 production data makes the
2026 to 2028 ramp contingent on a production scale-up that has not yet begun.
Tesla had guided a Gen 3 Optimus reveal for early 2026. As of May 2026, the reveal is
expected in late July or August 2026. External commercial sales remain guided for
"late 2026 at the earliest." Consumer sales: 2027. Every major product milestone
has slipped by at least one quarter. ARK's model does not appear to apply a probability
discount to milestones given this consistent pattern.
Figure AI deployed 40 F.02 robots at BMW's Spartanburg, South Carolina plant in January 2026.
The robots perform sheet metal positioning for welding operations, running 5 days per week
in 10-hour shifts. Over 10 months they contributed to the production of 30,000 BMW X3 vehicles
and moved more than 90,000 components with millimetre-accurate precision. Figure published
these metrics in its production report, the most detailed public operational disclosure
by any commercial humanoid robotics company. ARK Invest holds a venture stake in Figure
and has not issued public research updating its Optimus leadership narrative in light of
this deployment data.
§ 04 — The Hidden Portfolio
ARK's venture stakes in Tesla's competitors
ARK Venture Fund has invested in multiple humanoid robotics companies that compete directly
with Tesla's Optimus program. These investments are not disclosed in ARK's quarterly 13F-HR
filings, which only cover publicly traded securities. Retail investors reading ARK's public
humanoid robotics research have no visibility into these positions.
| Company |
ARK Investment |
Round / Date |
Estimated ARK Stake Value |
Deployment Status (May 2026) |
Mentioned in ARK Public Research |
| Figure AI |
Series B participation |
$675M · Feb 2024 |
~15× appreciated (Sep 2025 Series C at $39B) |
40 robots at BMW · 30,000 vehicles produced |
Minimal · mentioned once in Big Ideas 2026 |
| Apptronik |
Series A participation |
$403M · Mar 2025 (Google-led) + $520M · Feb 2026 |
Undisclosed |
Pilot deployments with NASA and GXO Logistics |
Not mentioned in major public research |
| 1X Technologies |
Investor |
Various rounds · OpenAI co-investor |
Undisclosed |
NEO Beta home robots · commercial testing |
Brief footnote in Big Ideas 2026 |
| Agility Robotics |
Investor |
Various rounds |
Undisclosed |
Digit robots in Amazon fulfillment pilots |
Named but not analyzed in public research |
| Tesla (TSLA) — public equity |
$1.31B in ARKK · dominant narrative focus · 0 commercial deployments |
Why this matters: ARK Venture Fund and ARK's public ETFs (ARKK, ARKQ) are
managed by the same firm under the same CIO. The venture fund's positions are not required
to appear in 13F filings. When ARK publishes research on humanoid robotics that elevates Tesla
while omitting Figure's BMW deployment data, readers cannot know that ARK has a financial
interest in both companies. ARK's research disclosures note general conflicts related to
Tesla's position size, but do not specifically disclose venture stakes in named competitors
when those competitors appear (or fail to appear) in research narrative.
The omission is structural, not incidental.
§ 05 — The Model vs. Reality
every assumption missed
ARK's open source Tesla valuation model for 2029 projects $1.23 trillion in revenue,
an $8.27 trillion enterprise value, and a 63.5% annual compound growth rate. The bull case
assumed robotaxi launch in 2025. It is mid-2026. Here is the model versus the record.
ARK model year-one assumptions vs. actual FY2025 results
Every major assumption missed
What this shows: ARK's model assumed a robotaxi launch in 2025, meaningful
Optimus production in 2025, and revenue growth consistent with a 63% CAGR. FY2025 actual
results: revenue essentially flat at $97.7B, net income down 46%, Optimus production at
hundreds vs. a target of 5,000 units, and no robotaxi in commercial service. The model's
2029 targets depend on a compound of still-to-materialize milestones. ARK has not published
a revised model addressing the 2025 misses.
| ARK Model Assumption |
ARK Projected (2025) |
Actual (FY2025) |
Status |
| Robotaxi launch (bull case) |
2025 |
Not launched |
Missed · now guided 2026 at earliest |
| Optimus units produced |
Meaningful production begins |
~hundreds (internal only) |
Missed · 90%+ below stated target |
| Revenue trajectory |
Growth consistent with 63% CAGR |
$97.7B · essentially flat |
Missed · no growth |
| Net income |
Expansion toward CAGR path |
$3.79B · down 46% |
Missed · 75% cumulative decline since FY2023 |
| Deliveries |
Growth trajectory |
1,636,129 · down 8.6% |
Second consecutive annual decline |
| FSD active subscriptions |
Rapid expansion |
Not separately disclosed by Tesla |
Unverifiable against model |
| Prior ARK Tesla target ($4,000 · 2024) |
$4,000/share by 2024 |
~$450 actual (missed by 89%) |
Missed |
The 2029 math still requires 39× EBITDA growth
ARK's 2029 Tesla bear case is $2,020 per share. The base case is $2,617. Both require
Tesla's enterprise value to reach roughly $8.27 trillion, approximately 29% of 2025 US GDP.
Tesla's FY2025 EBITDA was approximately $10.3 billion. To reach ARK's $400 billion EBITDA
milestone (which is also a condition of the CEO's November 2025 Performance Award),
Tesla would need to grow EBITDA 39-fold in four years. That rate of growth has never been
achieved by a company of Tesla's scale in the history of public markets.
ARK does not model the probability that this growth does not happen. Its published bear case
implies a 348% return from Tesla's Q4 2025 year-end price of approximately $451, and a 409%
return from the May 2026 price of $396.64. ARK's published work contains no scenario where
Tesla's current valuation is correct, let alone where it is too high.
§ 06 — The Structural Conflict
the $1.3B reason to be bullish
No law prohibits a fund manager from publishing bullish research on its largest holding.
No regulation requires that published humanoid robotics research disclose venture stakes
in the competitors being omitted. The conflict is structural and legal. That does not make
it invisible.
What a conflict-free analysis would require
A research firm without a financial stake in the outcome would evaluate Tesla Optimus
and Figure AI using the same framework: deployment count, commercial customer revenue,
operating cost per hour, ramp rate, and total addressable market capture.
On those metrics, Figure leads in every category where data exists. Tesla leads only
in capital committed and CEO attention, neither of which produces robot-hours logged
or customer contracts signed.
What ARK's conflict produces
ARK holds $1.31 billion in Tesla stock and earns approximately $10 million annually
in management fees from that position alone. ARK's Figure venture stake has appreciated
approximately 15x but generates no current fee income from an ETF. ARK's public narrative
elevates Tesla and mutes Figure.
The simplest explanation: the narrative reflects the incentive structure. This does
not require intent. It requires only that an organization with a large financial stake
in a conclusion tends to reach that conclusion.
ARK Tesla price targets published vs. prevailing Tesla stock price — 2020 to 2026
Targets always above actual price · by design
What this shows: ARK's published Tesla price targets (orange dots/line)
have consistently been set 3 to 10 times above the prevailing stock price at the time of publication.
The target always moves forward in time when missed. The stock never reaches the target on
the stated timeline. The research that supports these targets does not apply meaningful
probability to the scenario where the targets are wrong.
§ 07 — Anomalies
10 findings from the record
These anomalies are drawn from ARK's 13F-HR filings, published research documents, fund prospectuses,
and publicly available deployment data from humanoid robotics companies. Primary sources are cited
for each finding.
ARK's public research on humanoid robotics uses Tesla's Optimus development roadmap
as the primary organizing framework for the market. The research cites Tesla's AI stack,
Dojo supercomputer, and manufacturing scale as structural advantages. ARK simultaneously
holds $1.31 billion in Tesla stock and earns management fees on that position.
The research framework matches the firm's financial incentive
with precision that does not appear in competitors' research.
Figure published operational data from BMW Spartanburg: 40 robots, 30,000 vehicles,
90,000 components, 1.25 million steps, 1,250 operating hours. This is the most detailed
public operational dataset for any commercial humanoid robot deployment. ARK holds a
venture stake in Figure that appreciated approximately 15x in 2025. ARK has not
updated its published humanoid research to incorporate Figure's deployment data or
address how Figure's demonstrated performance compares to Optimus.
The most significant real world data point in ARK's own
portfolio does not appear in ARK's market narrative.
ARK's published Tesla model has a bear case of $2,020 per share by 2029. Tesla closed
Q4 2025 at approximately $451 (implied by ARK's reported position value of $1.31B across
2.9 million shares). The bear case implies a 348% return from that price, and a 409%
return from the May 2026 price of $396.64. A model where the worst-case scenario is
still 4 to 5 times the current price is not an analysis that accounts for downside risk.
It is a model built to justify bullish conclusions. No ARK Tesla publication addresses
the scenario where Tesla's 2029 enterprise value is at or below its 2025 level.
Some runs in ARK's published Monte Carlo simulation assume Tesla delivers 43 million
vehicles per year by 2029. Global passenger vehicle sales in 2025 totaled approximately
85 million units annually from all manufacturers combined. The most extreme tail scenarios
in ARK's model assume Tesla captures more than half the global automotive market in four
years from a base of 1.6 million deliveries. This assumption is not flagged as extreme
in ARK's published documentation.
ARK's N-2 filing for ARK Venture Fund discloses that Cathie Wood personally controls
65.68% of the fund through an initial $1 million investment, while simultaneously
serving as CEO, CIO, and Trustee of ARK Invest. This gives Wood direct personal
financial exposure to the fund's venture positions, including Figure, Apptronik,
and 1X. Wood's personal financial interest in the venture
portfolio, which includes Tesla's humanoid competitors, is not disclosed in
ARK's public humanoid robotics research.
ARK predicted a $4,000 Tesla share price by 2024 (actual: approximately $450 per share,
a miss of 89%). ARK predicted a $2,000 per share target by 2027 and a robotaxi launch
in the bull case by 2025. FSD is not commercially deployed at SAE Level 4 in any
jurisdiction. Optimus has no commercial customers. ARK's current model moves the
$2,617 target to 2029 without addressing why prior targets with similar assumptions
were missed. The pattern of extending targets rather than revising methodology is
consistent across five years of ARK Tesla research.
ARK's AUM peaked at $53.74 billion in Q2 2021, the period when its Tesla thesis
attracted the largest retail inflows. By Q4 2025, AUM had fallen to approximately
$15.07 billion. Tesla's stock price peaked in late 2021 and declined significantly
through 2022 and 2023 before partially recovering. ARK's AUM trajectory mirrors
Tesla's stock closely. The firm's business model depends on AUM recovery.
AUM recovery depends on Tesla's stock recovering. Tesla's stock recovering depends
partly on ARK's bullish narrative attracting new inflows. The circular dependency
is structural.
ARK's base case Tesla enterprise value for 2029 is $8.27 trillion. The US GDP in 2025
was approximately $29 trillion. ARK's target therefore requires Tesla to equal
approximately 29% of current US GDP in market capitalization. No company in history
has reached an enterprise value of this magnitude relative to its home country's GDP.
Apple, the largest company by market cap as of 2025, represents approximately 12% of
US GDP. ARK's target requires Tesla to reach approximately 2.4× Apple's current size.
ARK's Big Ideas research documents cite ARK's own prior research and valuation models
in their Works Cited sections. The $2.5 trillion SpaceX valuation cited in ARK's
2026 research appears as a Works Cited entry sourced to ARK's own earlier valuation work.
A research methodology that uses its own prior conclusions as supporting citations
cannot be independently validated. External citations to independent research on
robot deployment costs, market sizing, or competitor capabilities are sparse or absent
in the humanoid sections reviewed.
ARK's model assumes Optimus achieves human-level task performance by 2028 and
begins generating material revenue by 2028 to 2029. The model contains no scenario
where Optimus misses 2028 targets by the same margin it missed 2023, 2024, and 2025
targets. Given the complete miss on all prior Optimus milestones, a model that takes
the historical record seriously would assign meaningful probability to another delay.
ARK's model assigns near-zero probability to this outcome.
§ 08 — Open Questions
what ARK's research doesn't answer
These are questions for ARK Invest, not accusations. They are the questions any independent
research consumer should ask before relying on ARK's humanoid robotics analysis.
Venture disclosure
Why does ARK's humanoid robotics research not disclose ARK Venture Fund's positions in Figure, Apptronik, and 1X when those companies are discussed or omitted from analysis?
Figure AI update
ARK holds a venture stake in Figure AI, which has now deployed robots in BMW production generating documented commercial output. Will ARK publish updated research incorporating Figure's deployment data?
Bear case methodology
Why does ARK's Tesla bear case imply a 409% return from May 2026 prices? What scenario in ARK's model produces a negative or flat return for Tesla shareholders?
Target revision policy
ARK missed its $4,000 by 2024 target by 89%. It now has a $2,617 by 2029 target. What criteria would cause ARK to lower or withdraw a price target rather than extend the timeline?
Deployment comparison
Figure AI's F.02 has 1,250 documented operating hours in live factory production. Tesla's Optimus has zero commercial deployments. How does ARK's model weight current deployment status vs. future projections?
Cathie Wood's personal stake
Wood personally controls 65.68% of ARK Venture Fund, which holds Figure, Apptronik, and 1X. When Wood discusses humanoid robotics publicly, which of her financial interests is she representing?
Conflict disclosure standard
ARK discloses that it holds Tesla and has an incentive to show positive results. Does ARK apply the same disclosure when its research omits companies in which it holds undisclosed venture stakes?
The 43M delivery scenario
ARK's Monte Carlo includes tail scenarios where Tesla delivers 43 million vehicles per year. The global auto market is approximately 85 million units per year total. What is the methodology justifying this tail?
§ 09 — Sources
primary record
All factual claims in this investigation are sourced from primary documents: SEC filings,
ARK's own published research and models, company press releases, and earnings call transcripts.
ARK 13F-HR filings — Q1 2020 through Q4 2025
Quarterly institutional holdings reports disclosing ARK's public equity positions including Tesla share count and estimated market value. Filed with SEC.
EDGAR · CIK 0001779895 · 37 quarterly filings
ARK open source Tesla valuation model
ARK's publicly published Monte Carlo simulation projecting $2,617/share (base) and $2,020 (bear) by 2029. Source for enterprise value, revenue, and delivery assumptions.
ark-invest.com · accessed May 2026
ARK Big Ideas 2025 / 2026
ARK's annual flagship research reports. Source for humanoid robotics market sizing ($26T), competitor mentions, and Tesla advantage framing.
ark-invest.com · 2025–2026 editions
ARK Venture Fund N-2 filing
Registration statement for ARK Venture Fund disclosing Cathie Wood's 65.68% personal ownership through initial $1M investment, while serving as CEO, CIO, and Trustee.
SEC EDGAR · N-2 filing · ARK Venture Fund
Figure AI — BMW Spartanburg production report
Figure's published operational data: 40 F.02 robots, 30,000 BMW X3 vehicles, 90,000+ components, 1,250 operating hours, 1.2M steps, ~$25/robot-hour operating cost.
figure.ai · BMW Group press releases · Jan–Nov 2025
Tesla Q4 2025 earnings call transcript
Source for Musk's statement that Optimus units are deployed "primarily for learning, not productive tasks" and the several-hundred unit count.
Tesla investor relations · Q4 2025
Tesla FY2025 10-K
Annual report filed January 29, 2026. Source for revenue ($97.69B), net income ($3.79B), delivery count (1,636,129), and automotive gross margin (17.8%).
EDGAR · CIK 0001318605
Figure AI Series B and C filings / press releases
Figure's $675M Series B (February 2024) and $1B Series C (September 2025, $39B valuation). Source for ARK Venture Fund participation and appreciation estimates.
Figure AI press releases · Sacra research · 2024–2025
Methodology: This investigation does not allege that ARK Invest has violated
any law or regulation. ARK's fund disclosures acknowledge conflicts of interest related to
its Tesla position. The analysis presented here documents the gap between ARK's disclosed
financial positions, its published research narrative, and the available public data on
humanoid robot deployment as of May 18, 2026. All ARK 13F position data is drawn from
quarterly 13F-HR filings. ARK's price targets and model assumptions are taken from
ARK's own published documentation. Deployment data for Figure AI is taken from
Figure's own published production report and BMW Group press materials.
Not investment advice.