Too Big to Blitz? The Gate Failures Inside ARK's Funds

Twelve companies in ARK's published funds never receive a Blitz Index score, because they fail one of the two gates. Nine of them fail the same one — and most of those nine are among the most successful companies in the world. Here is why that is consistent, not contradictory.

Two gates before any points

The Blitz Index does not score every company it looks at. Two gates must both pass first:

• G1 — the business is real. Evidence of product-market fit: real, reordering product customers (for mature companies, sustained growth or a second growth curve).
• G2 — there is room. A ten-year market ceiling at least 10× current revenue.

A company that fails either gate gets a judgment file instead of a score. The ARK Lens marks these holdings "Gate failed" and excludes them from every average.

The gate failures ARK owns

In ARKF the largest holding in the fund is a gate failure. In ARKK, gate failures together outweigh Tesla, the fund's largest position.

G2: why the ceiling rises with success

The G2 bar is mechanical: ten times current revenue. That makes it a moving target that grows with the company. The scorecards already on BlitzLab show how quickly it scales:

Bars as stated on each company's October 2026 scorecard.

Alphabet's case is the instructive one. Even with four large markets combined, the bar is cleared rather than overwhelmed. For a company with several hundred billion dollars of annual revenue and a core market that is already largely captured, the arithmetic of a ten-fold ceiling becomes very hard to meet — and the gate is designed to say so.

That arithmetic is the natural reading of the mature names on ARK's G2 list — large incumbents in commerce and cloud, foundry, semiconductor test, pharmaceuticals and nuclear components — though each company's judgment file sets out its own reasoning. A G2 failure does not say these are weak businesses. It says the blitzscaling question — can concentrated capital still seize a category ten times larger than today's revenue? — no longer fits them. The ruler was built for companies with that kind of headroom left.

The crypto-adjacent G2 failures (Bitmine, Bullish) and Block sit in a different place: the question there is less about the size of the company than about how the gate reads the relevant market. Each case has its own judgment file; the Lens records only the verdict.

G1: where the line falls in gene editing

The three G1 failures are a different story: not too big, but not yet proven. Two are gene-editing companies held in ARKK, and the line between them and a scored peer is precise.

CRISPR Therapeutics passed G1 — borderline — because CASGEVY is approved, patients are being treated, and the joint venture's revenue is ramping, even though CRISPR's own income statement shows $0 of product revenue and every commercial function runs through Vertex. The gate accepted an approved product with paying patients and a 40% economic share as product-market fit routed through a partner.

Beam and Intellia are recorded at G1. By the gate's definition, their judgment files did not find real, reordering product customers. That is a statement about evidence on the scoring date, not about the science.

Archer Aviation's G1 verdict is best read next to its closest comparable in our files. In the Frontier Batch, Joby was rejected at G1 because it had no commercial service revenue and no quantified backlog for the aircraft itself. For Joby, the documented re-examination triggers are a type certificate, a quantified backlog in filings, or a first paid commercial flight — the kind of evidence that reopens any G1 verdict.

What gate weight does to the averages

Because gate failures are excluded, a fund's weight-averaged score describes only its blitz-eligible holdings. Two funds with the same average can differ sharply in how much of their capital sits outside the ruler's scope. ARKF's 65.7 is computed on 62.9% of the fund; 18.7% sits in gate failures, including its largest position.

This is intended. Assigning a score to a company the gates reject would mean inventing points the filings do not support. But it means a reader should look at gate weight alongside the average, not instead of it.

The other three funds: the gates bite harder

The table above covers the three funds published when this article was written. Since then the Lens has published the other three, and the gate picture there is heavier, not lighter. On holdings of 11/06/2026, gate-failed weight is 25.1% in ARKQ, 25.0% in ARKX and 27.4% in ARKG — each above the highest of the first three funds, and each above a quarter of the fund. The number of distinct gate-failed tickers held across all six funds is 31: 16 at G1 and 15 at G2.

The three new funds fail in opposite directions, and the contrast is the finding. ARKQ and ARKX are G2 funds: their gate weight is dominated by large industrial and supplier names — Deere (3.51% in ARKQ, 6.51% in ARKX), Caterpillar, Lockheed Martin, Teradyne, Intuitive Surgical and Synopsys — companies whose revenues are already too large for a ten-fold market ceiling, in the same way Amazon and TSMC fail it in the first three funds. ARKQ’s G1 weight, by contrast, is small (5.5%) and concentrated in aviation: Archer at 2.81%, Joby at 1.90% and Oklo at 0.78%.

ARKG is the mirror image. Only one of its thirteen gate failures is at G2 — Eli Lilly at 4.40% — and the other twelve, worth 23.0% of the fund, fail G1. Absci alone is 5.31% of ARKG, the largest single gate failure in any of the six funds; Beam (4.15%), Compass (2.29%), Nurix (2.06%) and Intellia (1.95%) follow. Genomics is where the proof gate bites hardest: the science is pre-commercial, so the ruler’s demand for real, reordering product customers excludes nearly a quarter of the fund. A reader who compares ARKG’s average with ARKQ’s without looking at this table is comparing two different kinds of absence — one fund’s unscored weight is mostly companies that are too big, the other’s is mostly companies that are not yet proven.

The same widening shows up in how much of each fund the ruler can see at all. On 11/06/2026, scored holdings account for 75.2% of ARKK, 74.3% of ARKW and 64.7% of ARKF, against 69.2% of ARKQ, 65.2% of ARKX and 70.7% of ARKG. The remainder in each fund is gate failures plus unscored and private positions. In the three newer funds, roughly a third of the fund sits outside the ruler’s scope — a larger blind spot than in ARKK or ARKW, and a reason to read their averages with the gate table open beside them.

Two different kinds of absence

It helps to keep the two gates’ meanings apart, because they fail for opposite reasons and reopen on opposite evidence. A G2 failure is a statement about arithmetic: the ten-year market ceiling does not reach ten times current revenue. The company can be superb — the list includes the world’s largest foundry, its largest e-commerce company and a leading test-equipment maker — and the verdict is unchanged, because the gate asks about headroom, not quality. What would reopen a G2 verdict is a change in the arithmetic: a larger addressable market, a new market the company can credibly enter, or a revenue mix that resets the bar. Alphabet’s scorecard shows how narrow that path is even for a company that passes: four large markets combined only just clear a ~$4.46T bar.

A G1 failure is a statement about evidence: no real, reordering product customers were found on the scoring date. It is the more reversible of the two, because the evidence it waits for is concrete — an approved product with paying patients, a first paid commercial flight, a quantified backlog in filings. Joby’s judgment file names exactly those triggers. The gene-editing line drawn in this article is the same kind of line: CRISPR passed, borderline, because CASGEVY is approved and patients are being treated; Beam and Intellia did not, because their judgment files found no equivalent evidence. A G1 verdict therefore says less about the future than a G2 verdict does. It says “not yet proven”, where G2 says “no longer room” — and the funds reflect that: G1 weight is concentrated in the youngest themes (genomics, aviation), G2 weight in the oldest (commerce, industrials, suppliers).

One more distinction matters for readers of the averages. Gate failures are excluded from every Lens aggregate, so a fund with 27% gate weight and a fund with 15% gate weight can report similar averages while describing very different fractions of their capital. ARKF’s 65.7, computed on 62.9% of the fund, is the sharpest example in the first three funds; ARKG’s average is computed on a similar basis, with 27.4% of the fund in gate failures. The average is not wrong. It is partial — and the gate table is the measure of how partial.

The boundary

A gate failure is a judgment about whether the blitzscaling ruler applies, not a quality grade and not an opinion on the stock. Several of the companies above are among the most profitable businesses in their industries. Gates are re-examined when the evidence named in the judgment file appears.

BlitzLab scores businesses against a scholar's standard — not investment advice. No price targets, no buy/sell recommendations. Holdings come from ARK Invest's daily public disclosures; BlitzLab is not affiliated with ARK Invest and this research is not endorsed by them.

Sources

1. BlitzLab, ARK Lens fund pages (gate status per holding) — https://blitzlab.app/ark/arkk/ · https://blitzlab.app/ark/arkw/ · https://blitzlab.app/ark/arkf/ — holdings as of 11/04/2026 — checked 2026-11-05
2. BlitzLab, "About BlitzLab" (gate definitions) — https://blitzlab.app/about/ — checked 2026-11-05
3. BlitzLab scorecards (G2 bars): Circle https://blitzlab.app/companies/crcl/ · Coinbase https://blitzlab.app/companies/coin/ · Tesla https://blitzlab.app/companies/tesla/ · Alphabet https://blitzlab.app/companies/googl/ — Oct 2026 — checked 2026-11-05
4. BlitzLab, "CRISPR Therapeutics — Blitz Index 48 (C)" (G1 borderline pass) — https://blitzlab.app/companies/crsp/ — checked 2026-11-05
5. BlitzLab, "The Frontier Batch: Four Scores, Two Gate Rejections" (Joby G1) — https://blitzlab.app/research/frontier — 2026-10-15 — checked 2026-11-05