What It Actually Costs to Launch an ETF

Costs here come from provider and exchange schedules published between 2023 and September 2026. Each cost table names its own source and date.

The paperwork to get an ETF listed is cheaper than most people assume. Two white-label providers publish their prices, and both put a straightforward launch near $50,000 to $75,000 of one-time cost, with a first trade three to five months after you start.¹³ A Nasdaq listing carries no application fee and $4,000 a year.⁸

The recurring cost is the part that decides everything. Operating the registered investment company that sits behind the ticker runs roughly $175,000 to $300,000 a year across the three published estimates, before anyone is paid to sell the fund.¹³⁴ On a 0.50% fee and $200,000 of fixed cost, with the platform also taking 10 basis points of assets, the fund needs roughly $50 million before the sponsor keeps a dollar. On a 0.25% fee the same cost stack needs about $133 million. Below roughly 15 basis points, on these assumptions, there is no level of assets that works at all.

That last point is the useful one, and it holds whether or not you would ever launch anything. The cost of running a fund sets a floor under what a small sponsor can charge, which is visible in the expense ratio of every niche ETF you have ever looked at, and it sets a level of assets below which a fund is being kept alive out of its sponsor’s pocket.

What sits behind the ticker

An ETF is a registered investment company. The share you buy is an interest in a fund that has a board, an adviser, a compliance program, a custodian, an administrator, a transfer agent, a principal underwriter, an auditor, an exchange listing, and a set of firms willing to create and redeem its shares. Rule 6c-11 removed the need for most ETFs to obtain an individual exemptive order from the SEC before launching, which is the single largest reason a small sponsor can attempt this at all.⁵

The disclosure row is more work than it sounds. Rule 6c-11 requires a qualifying ETF to post its portfolio holdings before the start of each trading day, with ticker, CUSIP, quantity and portfolio weight for every position, alongside prior-day NAV, market price, and premium or discount. It requires a premium and discount table and a line graph, a median bid-ask spread computed from the national best bid and offer sampled every ten seconds across thirty calendar days, and a separate disclosure held for a year whenever the premium or discount exceeds 2% for more than seven consecutive trading days. A fund using custom baskets needs written policies naming who reviews them.⁵ A white-label provider does all of this. It is a large part of what the annual fee buys.

A note on what 6c-11 does and does not cover

The SEC’s small-entity compliance guide for Rule 6c-11 is dated December 2019 and lists leveraged and inverse ETFs among the funds excluded from the rule, which is no longer accurate. When the Commission adopted Rule 18f-4 in late 2020 it amended 6c-11 to bring leveraged and inverse ETFs inside the rule, provided they comply with 18f-4.⁶ Unit investment trusts, share-class ETFs, and non-transparent active ETFs remain outside it.

Why this matters if you only ever buy ETFs

Four things follow from the cost structure, and all four are checkable before you buy.

A fund below its break-even is being subsidized

Sponsors do not disclose a break-even, but you can estimate one. If a fund charges 0.50% and holds $20 million, it generates $100,000 a year of gross fee against an operating stack that starts around $175,000. The gap is coming out of the sponsor’s balance sheet, and it is coming out for as long as the sponsor is willing. Morningstar assumes $250,000 of annual fixed cost for a representative active ETF and puts break-even near $33 million against the 0.75% average net expense ratio of that group.¹³

This is closure risk expressed as arithmetic rather than as a feeling about a sponsor. It is also why the closure statistics look the way they do, which is the next section.

Why a small sponsor cannot offer a three-basis-point fund

The platform’s marginal charge, quoted at 5 to 15 basis points depending on scale, comes off the top of the management fee.² What is left has to cover the fixed cost. A sponsor charging 10 basis points and paying 10 basis points of marginal cost retains nothing per incremental dollar, so no level of assets ever covers the fixed stack. This is why a three-basis-point total market fund comes from Vanguard, BlackRock or State Street and not from a two-person shop with a good idea. The three-basis-point fund is possible only for a sponsor that owns its own infrastructure and spreads it across trillions.

When you see a thematic or factor ETF at 59 or 75 or 95 basis points, at least part of that is a cost floor rather than a judgment about what the strategy is worth.

A fee waiver is a subsidy that can reverse

Many small funds quote a net expense ratio held down by an expense limitation agreement with a stated end date. These agreements commonly let the adviser recover what it gave up. The Logan Capital Large Cap Growth ETF is a representative example: the adviser “may request recoupment of previously waived fees and paid expenses from the Fund for 36 months from the date they were waived or paid,” capped at the lesser of the expense cap in place when the waiver happened or the one in place at recoupment.²⁰

The quoted fee on a small fund is therefore a fee for a stated period, with a mechanism attached for raising it later. The agreement and its expiry date are in the prospectus.

Low assets signal closure risk more than trading cost

Low assets are a real signal about whether a fund will still exist in three years. They are a much weaker signal about whether you can trade it today. Fidelity describes the two layers plainly: “ETFs have 2 layers of liquidity: liquidity of the underlying securities, i.e., the primary market, and the available liquidity in the secondary market,” and notes that in the primary market an authorized participant “can change the supply of ETF shares available.”¹⁸

A small fund holding large-cap US equities has a creation basket a market maker can hedge and assemble cheaply, so its quoted spread can be tight even on thin volume. A large fund holding illiquid credit can be harder to move. Reading a fund’s assets as a proxy for its tradability conflates two separate questions. The inference here is ours rather than Fidelity’s wording, but it follows directly from the two-layer structure that both Fidelity and the SEC describe.

The three routes, and what each costs

There are three ways to bring a fund to market, and they differ by an order of magnitude. These are published estimates from two providers and one consultancy, each as of its own date.

ETF Architect quotes the white-label route at start-up costs of “roughly $50k and ongoing all-in costs of $175k to $225k per year. Plus, additional marginal costs vary from 5 bps to 15 bps depending on the scale of the fund,” and says a launch “can occur as quickly as 90 days.”¹ Its published breakdown of the start-up figure is $8,000 to $20,000 of platform launch work, $22,500 to $30,000 for the registration statement, $3,000 to $5,000 of SEC and XBRL filings, and $15,000 to $20,000 of board preparation, with complicated cases running from $40,000 to $150,000 and up.²

ACA Group, writing as a consultant rather than a platform, put series trust operating expenses at “at least $200 – $300K to operate the ETF in a Series Trust model” and the “break-even point to be $30 – $50M, as well as having the ability to bring their own assets.”⁴ Three independent estimates landing in the same band is the strongest evidence available that the band is real.

Why the timeline is months rather than years

Two mechanisms compress it. Adding a series to an existing registered open-end fund becomes effective on the seventy-fifth day after filing under Rule 485(a)(2), without the SEC having to affirmatively act.⁷ And a plain ETF lists under an exchange’s generic listing standards, which set objective criteria a fund can simply meet, rather than requiring a fund-specific Form 19b-4 rule filing with its own comment period and SEC review. The relevant rules are NYSE Arca Rule 8.600-E, Nasdaq Rule 5735, and Cboe BZX Rule 14.11.⁹

Neither mechanism helps a fund that needs its own exchange rule change. That is the difference between a launch measured in months and one measured in years, and it is the main reason a conventional equity ETF is a fundamentally different project from a novel one.

Listing on an exchange is the cheap part

Listing fees are close to a rounding error against the operating stack, though they are not identical across venues.

Cboe’s annual fee falls as volume rises, which is worth noting because it runs opposite to intuition. Nasdaq added an optional “Premier” annual fee of $50,000 per product in September 2026, buying eligibility for a liquidity provider program; the standard fee stayed at $4,000.⁸ One operational detail for a fund at the end of its life: as of February 2026 Cboe no longer prorates refunds of the annual fee when an ETP liquidates, while Arca still does.¹⁰

Seed capital buys shares you still own

This distinction gets lost constantly, and it changes the shape of the decision. If you put $5 million into your own fund, you own $5 million of ETF shares. The money is committed and exposed to the strategy, and it is not spent. The $250,000 you pay the platform is spent.

There is no $5 million requirement anywhere in the rules. The actual statutory floor is Section 14(a) of the Investment Company Act, which bars a registered fund organized after August 1940 from making a public offering unless it has net worth of at least $100,000, or has previously made such an offering, or has firm agreements from no more than 25 responsible persons totaling $100,000 and refundable if not reached within 90 days.¹¹ The provision has never been indexed for inflation in 86 years, which is why $100,000 keeps showing up in filings.

Real filings show both numbers in the same document. The Syntax Stratified LargeCap ETF filed an audited seed balance sheet of exactly $100,000, being 2,500 shares at $40.00, and described that as “10% of a Creation Unit,” with an additional $900,000 deposit making up a $1,000,000 creation unit.¹² ETF Architect puts the practical range at a “minimum seed… typically 10k -100k shares, typically priced at $25/share, so you are initially looking at a $250k-$2.5mm seed.”²

The larger figures you see quoted are commercial screens rather than legal minimums. ETF Architect says that “at a minimum, we expect partners to possess $500k in operating capital, $5M of launch capital for the ETF on Day 1, and a reasonable roadmap to $50M in fund AUM.”¹ That is an experienced operator stating what it considers commercially credible, which is more informative than a legal minimum precisely because it is a business judgment.

The break-even arithmetic

The version of this calculation that circulates divides fixed cost by the fee. That omits the marginal platform charge, and the omission grows as the fee falls. With F as fixed annual cost, f as the management fee in basis points and v as the marginal platform cost in basis points, break-even assets are:

break-even AUM = F ÷ ((f − v) ÷ 10,000)

The denominator drives the answer. On $200,000 of fixed cost and a 10 basis point marginal charge, a 0.50% fund breaks even near $50 million, while a 0.25% fund needs about $133 million rather than the $80 million the simpler formula gives. As f approaches v the required assets go to infinity, and at or below it there is no solution.

Run your own numbers

Set the fee, the marginal platform charge, and your own cost stack. The chart shows annual sponsor profit against assets at three fee levels; each line crosses zero at its own break-even. Drag the marginal cost slider up and the lines flatten toward horizontal, which is the fee floor appearing.

Where the unitary fee goes

Most white-label funds use a unitary fee, under which the adviser pays essentially all fund operating expenses out of one number. The Defiance MANGOS ETF prospectus is typical: the adviser “will pay, or require a sub-adviser to pay, all of the Fund’s expenses, except for” advisory fees, interest, dividends on securities sold short, taxes, brokerage, acquired fund fees, 12b-1 fees, litigation and extraordinary expenses. The fund’s total annual operating expenses are 0.51%, being a 0.49% management fee plus 0.02% of other expenses.¹⁹

So the 0.49% is the number out of which custody, fund accounting, transfer agency, audit, trustee fees and index licensing all come. On a $20 million fund that is about $98,000 of gross revenue against a service stack that starts higher. Reading a small fund’s expense ratio as the sponsor’s margin overstates the margin by most of itself.

Most closures happen below $25 million

Morningstar counted about 150 active ETF closures in 2025, precisely 146, split between 114 liquidations and 32 mergers. Only six of the closed funds held more than $50 million at the start of 2025, most held under $25 million, and most had existed for only about 1.75 years.¹⁴ At the end of 2025 roughly 1,260 active ETFs still held under $50 million.¹⁴ Set against an industry of 5,182 US ETFs holding $15.67 trillion as of July 2026, the failures are concentrated almost entirely in the small end.¹⁵

The mechanics of a closure are worth knowing because they are faster than most people expect. Two real 2026 examples, taken from the funds’ own filings:

Both gave about three weeks between the filing and the last day of trading, with cash following within a week. Themes told shareholders they “may sell their holdings of a Fund until market close on July 28, 2026” and that the resulting distributions “are taxable events… may include accrued capital gains and dividends.”¹⁶ KraneShares added the detail that matters most to anyone who does not act: between the closing date and liquidation, “shareholders may only be able to sell their shares to certain broker-dealers. There is no assurance that there will be a market.”¹⁷

In a taxable account this is a realization event on the fund’s schedule. If the position is at a gain, the gain arrives in the year the fund chose. In an IRA or 401(k) nothing is taxed and the cash simply lands in the account. The holder’s side of this, including what it means for a bespoke factor fund held at a large embedded gain, is covered in more depth in the guide on sponsor risk in taxable accounts.

When launching is actually reasonable

The binding question is where the first $30 million to $50 million comes from. Launch cost is small relative to the runway needed to reach that level, so the constraint is distribution rather than capital.

The candidates who clear that bar mostly share one feature: they already manage the assets, or they already control the distribution.

• An RIA or SMA manager with an existing strategy at scale. Day-one assets are already under management and the track record already exists. This is the strongest case, and it is why platforms court this group hardest.
• An established asset manager adding a wrapper. Distribution relationships exist; the ETF is a new delivery mechanism for a product that already sells.
• A research or index firm with a committed distribution partner. The commitment has to be contractual rather than enthusiastic.

An audience is not distribution. A large following converts to assets at rates that are usually far lower than the follower count suggests, and the fund still has to clear platform review. Exchange Traded Concepts, a white-label provider writing about its own funds’ experience, says that “six to twelve months of live trading history is standard before most platforms will consider a fund,” and that “being approved for a platform is distinct from being on a recommended or model list, which takes considerably longer and requires a sustained trading track record.”²¹ That is a vendor describing its own market, so treat it as informed rather than disinterested, but it points the right way: the fund has to survive its first year before the channels that could make it viable will look at it.

The case for doing it anyway

White-labeling has produced real businesses. YieldMax, built on Tidal infrastructure as a joint venture, runs a family of option-income funds of which the TSLA fund alone held roughly $669 million in September 2026.²⁵ The wrapper also delivers something an SMA cannot: in-kind creation and redemption gives an ETF structural tax efficiency, and a shared trust supplies board oversight and service relationships that would take years to assemble alone. For a manager who already has the assets and wants them in a better wrapper, the arithmetic above is affordable rather than prohibitive.

The Section 351 route, briefly

One development explains why platforms are courting RIAs so aggressively. Under IRC Section 351, an adviser can sometimes contribute appreciated taxable portfolios into a newly formed ETF without immediately realizing gain, which answers the day-one assets question without finding new investors. The transferor group must control at least 80% of the entity immediately after the transfer, and because the recipient is an investment company, the contributed portfolio must be diversified under the 25 and 50 percent tests of Section 368(a)(2)(F)(ii): no more than 25% of the value of total assets in any one issuer and no more than 50% in five or fewer issuers.²² The test applies per transferor, which is why these launches are syndicated across many contributing accounts.

The law here is unsettled. There is no regulation, revenue ruling, or published guidance on the technique, and issuers say so in their own risk factors: the Oakmark US Concentrated ETF discloses that “the IRS could seek to apply the step transaction doctrine… neither the Fund nor the Adviser has obtained a ruling from the IRS.”²³ Legislation has been introduced that would deny nonrecognition on transfers of marketable securities to regulated investment companies going forward. Treat it as tax-counsel territory. The fund-level detail is covered in the Alpha Architect guide, which looks at the funds built for this purpose.

One dated caveat

On June 30, 2026 the SEC issued a request for comment on novel ETFs, covering funds “seeking to invest in innovative asset classes or engage in novel investment strategies,” naming crypto assets, commodity-focused instruments, single-stock strategies, heightened leverage, private assets and event contracts. Comments closed August 31, 2026 and no rule proposal had issued as of September 2026.²⁴ Nothing in it disturbs the cost or timeline assumptions for a conventional transparent equity ETF, but any fund in those categories should be assumed to face a different process entirely.

When the numbers here do not apply

Every figure above describes a plain vanilla fund. They understate a launch that involves a novel asset class, a derivatives-heavy strategy requiring a Rule 18f-4 derivatives risk management program, a non-transparent active structure, or anything needing its own exchange rule filing. They also understate a fund with an expensive licensed index, since index licensing is a pass-through cost that can run into six figures on its own for a well-known benchmark.

In the other direction, they overstate the incremental cost of a second or third fund on a platform you already use. Much of the setup work and board time is shared, which is why sponsors who reach scale tend to launch families rather than single funds.

Key takeaways

• Creating the fund is cheap; operating it is not. Two providers put a white-label launch at $50,000 to $75,000 and three independent estimates put annual operations between $175,000 and $300,000.
• Break-even depends on the fee net of the marginal platform charge. Dividing fixed cost by the headline fee understates the required assets, badly so at low fees.
• There is a fee floor. At a management fee at or below the platform’s marginal take, no level of assets covers the fixed cost. This is why the cheapest index funds come only from the largest sponsors.
• Seed capital is not spending. The statutory floor is $100,000 under Section 14(a); the practical range providers quote is $250,000 to $2.5 million; $5 million is one platform’s commercial screen.
• Closures concentrate in small, young funds. Of about 146 active ETF closures in 2025, most held under $25 million and most were about 1.75 years old. Notice periods in two 2026 examples ran about three weeks.
• Assets signal survival odds; underlying holdings signal tradability. The two get conflated and they answer different questions.

How Summitward helps

Frequently asked questions

How much does it cost to launch an ETF?

On published white-label pricing read in September 2026, roughly $50,000 to $75,000 of one-time cost and $200,000 to $260,000 a year to operate, with ETF Architect quoting $175,000 to $225,000 annually plus 5 to 15 basis points of marginal cost. A credible first-year budget including marketing is larger, and the operating cost continues whether or not the fund gathers assets.

How much money do you need to start an ETF?

Legally, $100,000 of net worth before a public offering under Section 14(a) of the Investment Company Act. Commercially, one platform requires $500,000 of operating capital and $5 million of day-one fund assets before it will take a partner. The commercial screen binds long before the legal minimum does.

How much AUM does an ETF need to be profitable?

It depends on the fee and the cost stack, so it is a formula rather than a number. On $200,000 of fixed cost and a 10 basis point marginal charge, a 0.50% fund needs about $50 million and a 0.25% fund about $133 million. Morningstar puts a representative active ETF near $33 million against a 0.75% net expense ratio, and ACA Group cites $30 million to $50 million as the range providers work with.

Why do small ETFs charge so much?

Partly because they have to. A fund’s fixed operating cost does not shrink with its assets, and the platform takes a marginal share on top, so there is a fee below which no amount of assets covers the cost. A small sponsor cannot offer a three-basis-point fund at any size.

What happens if an ETF I own closes?

The fund files a notice, typically giving several weeks. In two 2026 examples the gap between filing and the last trading day was about three weeks. You can sell in the market until the last trading day. If you hold through it you receive a liquidating distribution in cash, which is a taxable event in a taxable account and may carry accrued capital gains and dividends. In an IRA or 401(k) it is not taxed.

Is a low-asset ETF hard to trade?

Not necessarily. An ETF has two liquidity layers, the secondary market where shares trade and the primary market where authorized participants create and redeem against the underlying securities. A small fund holding liquid large-cap equities can quote tightly on low volume. Low assets are a better signal about closure risk than about spreads.

Do you need SEC approval to launch an ETF?

Not an individual exemptive order, for a conventional fund. Rule 6c-11, effective since December 2019, lets qualifying ETFs operate without one. A new series added to an existing registered fund becomes effective on the seventy-fifth day after filing under Rule 485(a)(2), and a plain fund lists under an exchange’s generic listing standards without a fund-specific rule filing.

Related guides

• If Dimensional Gets Sold, Should Factor Investors Care?: sponsor risk and tax lock-in from the holder’s side.
• The Rise of Avantis: what it looks like when a new fund company clears the scale problem.
• Alpha Architect ETFs: QVAL, QMOM, and the Case for Concentrated Factors: the funds built to serve Section 351 conversions.
• BOXX Explained: the in-kind creation and redemption mechanism in detail.

Sources

1. ETF Architect, “How to Start an ETF? Resources and FAQ,” page modified August 28, 2026 (start-up and annual costs, marginal basis points, partner minimums, 90-day timeline). etfarchitect.com
2. ETF Architect, “ETF-preneurs: An Introduction to ETF White Label Services,” page modified August 28, 2026 (start-up cost breakdown, scale basis points, practical seed range). etfarchitect.com
3. Tidal Financial Group, “Launching an ETF? White Label Could Be the Answer,” November 12, 2024 (three-route cost and timeline comparison). tidalfinancialgroup.com
4. Joe Higgins, ACA Group, “A Primer on Launching an ETF Under a Turn-Key Structure,” April 28, 2023 (series trust operating expenses and break-even range). acaglobal.com
5. SEC, “Exchange-Traded Funds: A Small Entity Compliance Guide,” December 6, 2019, and Rule 6c-11, 17 CFR 270.6c-11 (conditions, daily website disclosure, custom basket policies). sec.gov
6. SEC, “Use of Derivatives by Registered Investment Companies and Business Development Companies,” Release No. IC-34084 (amending Rule 6c-11 to include leveraged and inverse ETFs subject to Rule 18f-4). sec.gov (PDF)
7. 17 CFR 230.485(a)(2), effective date of post-effective amendments adding a series (seventy-fifth day after filing). law.cornell.edu
8. Nasdaq ETP Listing Guide, September 2025, and SR-NASDAQ-2026-068 (Release 34-106213, 91 FR 56234), operative September 1, 2026 ($0 entry fee, $4,000 standard annual fee, optional $50,000 premier fee). nasdaq.com
9. NYSE Arca Rule 8.600-E, Nasdaq Rule 5735 and Cboe BZX Rule 14.11, generic listing standards for Managed Fund Shares and related products, as cited in registration and exchange filings on EDGAR. sec.gov
10. Cboe BZX ETP Listings Compliance Guide, April 22, 2026, and Rule 14.13(b)(2)(E)(vi) (first-year $4,500; annual fee by consolidated average daily volume). cboe.com
11. Investment Company Act of 1940 Section 14(a), 15 U.S.C. 80a-14(a) ($100,000 net worth before public offering). law.cornell.edu
12. Syntax Stratified LargeCap ETF, Form 485BPOS filed April 25, 2019 (audited seed balance sheet of $100,000, 2,500 shares at $40.00, and creation unit description). sec.gov
13. Stephen Welch, Morningstar, “Why Some Active ETFs Fail in 9 Charts,” February 9, 2026 ($250,000 assumed annual fixed cost; $33 million break-even against a 75 basis point net expense ratio). morningstar.com
14. Morningstar Manager Research, “Active ETF Launches and Closures: 2025 in Review,” February 13, 2026 (146 closures, asset and age distribution, funds under $50 million). morningstar.com
15. Investment Company Institute, “Exchange-Traded Fund Data, July 2026,” released August 28, 2026 (5,182 ETFs, $15.67 trillion). ici.org
16. Themes ETF Trust, Form 497 filed July 6, 2026 (liquidation of CZAR, HWAY and USRD; cease trading July 28, 2026; distribution on or about July 31, 2026). sec.gov
17. KraneShares Trust, Form 497 filed February 20, 2026 (liquidation of KLXY and KEUA; last trading March 13, 2026; liquidation March 20, 2026). sec.gov
18. Fidelity, “ETF Liquidity,” Learning Center (two layers of ETF liquidity; role of authorized participants). fidelity.com
19. Defiance MANGOS ETF, Tidal Trust II, Form 485BPOS filed September 4, 2026 (unitary fee language; 0.49% management fee, 0.51% total expenses). sec.gov
20. Logan Capital Large Cap Growth ETF, Advisors Series Trust, Form 485BPOS filed August 28, 2026 (36-month recoupment of waived fees, subject to the lesser of two expense caps). sec.gov
21. Exchange Traded Concepts, “Your ETF Is Live. Now the Real Liquidity Work Begins,” June 15, 2026 (platform review expectations; a white-label provider writing about its own market). exchangetradedconcepts.com
22. 26 U.S.C. 368(a)(2)(F)(ii) (25 and 50 percent diversification tests by value of total assets), applied to Section 351 transfers by Treas. Reg. 1.351-1(c)(6)(i); 26 U.S.C. 351(e)(1) and 368(c). law.cornell.edu
23. Oakmark U.S. Concentrated ETF, Form 497 filed June 25, 2026 (Code Section 351 tax recharacterization risk factor; step transaction doctrine; no IRS ruling obtained). sec.gov
24. SEC, “Request for Comment on Novel Exchange-Traded Products,” Release Nos. 33-11426; 34-105808; IC-36228, File No. S7-2026-24, issued June 30, 2026, 91 FR 40647; comments due August 31, 2026. sec.gov
25. YieldMax TSLA Option Income Strategy ETF (TSLY) net assets of approximately $669 million as of September 4, 2026; YieldMax trademark jointly owned by Tidal Investments LLC, ZEGA Financial LLC, Lucania Investments LLC and Level ETF Ventures LLC per the issuer’s site. yieldmaxetfs.com

Author disclosure

I have never launched an ETF and have no commercial relationship with any provider named here. Provider figures are their own published list prices on the dates given, gathered from public pages, and none of them were contacted for this article. Nothing here is legal, tax or investment advice; the Section 351 discussion in particular is a summary of unsettled law.