Bitcoin ETF fees: the expense ratio is only a third of the cost

Sponsor fees, cheapest first

Start with the number everyone quotes. The sponsor fee is an annual percentage of net assets, accrued daily and taken out of the fund rather than billed to you. You never see it leave your account; it simply means each share represents slightly less bitcoin at the end of the year than it did at the start. That is why it is easy to ignore and why it compounds quietly against you.

The dollar columns below are deliberately blunt. A percentage feels abstract. Sixty-five dollars a year against seven hundred and fifty on the same $50,000 does not.

Two observations before we move on. The cheapest fund on the list, Morgan Stanley's MSBT at 0.14%, is also the newest — it began trading on 8 April 2026 — and there is no reliable published figure for its assets. Cheap is not automatically better when a fund is small and young. The second-cheapest, Grayscale's Mini Trust at 0.15%, has both scale and history, and for most buyers it is the more sensible interpretation of "cheapest at scale". The full roster, with assets and coin counts, sits on the Bitcoin ETF list.

Three costs, not one

Here is the argument this page exists to make. A spot Bitcoin ETF charges you in three separate ways, and only one of them appears in the fact sheet.

Sponsor fee

An annual percentage of net assets, accrued daily. Paid by every holder, every day, regardless of whether they trade. Disclosed, comparable, and the one number that gets quoted.

Bid-ask spread

The gap between what a market maker will buy at and what it will sell at. You pay it on the way in and again on the way out. Disclosed nowhere in the expense ratio, and it scales with how often you trade, not with how long you hold.

Tracking difference

The gap between the fund's return and bitcoin's return over a period, after fees. It captures execution quality, cash drag, creation and redemption friction and index construction. It is measured after the fact, so nobody can promise you a number in advance.

These three costs hit different investors differently, which is why there is no single cheapest fund. Someone who buys once and holds for a decade pays the spread twice and the sponsor fee three thousand six hundred and fifty times. Someone who contributes every payday pays the spread constantly, and for them a fund that is two basis points cheaper but five basis points wider is a bad trade.

Cost is three numbers

The bid-ask spread you cross

A spread is not a fee anyone charges you. It is the price of immediacy — the compensation a market maker takes for standing ready to trade with you at a moment's notice. On the largest Bitcoin ETFs it is remarkably small. IBIT has quoted a median spread of about 0.03% and has traded in penny increments since 2 January 2025, when it entered the exchange's Penny Interval Program. On a $500 order, three basis points is fifteen cents.

The most interesting row in the table below is GBTC. It is the most expensive fund in the category by a wide margin and it prints the tightest quote — roughly 0.02% on 1 September 2026. That is not a contradiction. Spread reflects the depth and competitiveness of the market making around a fund, not its fee, and GBTC has an enormous shareholder base constantly trading against a deep options market. A tight spread does not make an expensive fund cheap. It just means the entry ticket is cheap.

Work the arithmetic through

Take someone contributing $500 a month. Over a year that is twelve purchases and $6,000 deployed. Suppose the fund they chose quotes 0.11% rather than 0.03% — the gap between the smallest fund in the category and the largest. The extra eight basis points cost them forty cents on each $500 order, so $4.80 across the year.

Now compare that to an expense ratio. Eight basis points on a $6,000 balance is also $4.80. In the first year of a monthly contribution plan, a 0.08 percentage point wider spread costs about the same as a 0.08 percentage point higher sponsor fee. That is the equivalence worth carrying around: at that turnover, spread and expense ratio are the same currency.

The equivalence decays, and it is worth being honest about how. Spread cost scales with what you contribute this year; fee cost scales with everything you have accumulated. By year ten of the same plan the balance is roughly ten times the annual contribution, so the sponsor fee is doing roughly ten times the damage the spread is. The practical rule falls out of that arithmetic on its own: if you are building a position, weight liquidity; once the position is built, weight the fee. Buy-and-hold investors can more or less ignore the spread column. People who contribute every month cannot.

None of this argues for market orders. Whatever fund you pick, a limit order costs nothing and protects you on the days when the quoted spread is nothing like the median — see our guide to ETF order types for the mechanics.

Tracking difference: the cost nobody advertises

If a fund charges 0.19% and bitcoin returns 40% over a year, you would expect roughly 39.81%. Tracking difference is the gap between that expectation and what actually landed. In a well-run spot fund the gap should be close to the fee and nothing more. In practice it is not always.

Third-party estimates put IBIT and FBTC around 0.03%, Bitwise's BITB around 0.15%, and Franklin's EZBC around 0.42% — the worst of the majors, and materially worse than its 0.19% headline fee would suggest. If that figure holds, EZBC's fee advantage over IBIT is not merely erased; it is reversed several times over.

One structural point helps explain why the largest funds track best. Since the SEC permitted in-kind creations and redemptions on 29 July 2025, authorised participants can deliver bitcoin directly rather than cash, which removes a conversion leg and the slippage attached to it at fund level. The funds with the deepest AP relationships got the most out of that change. We cover who those participants are and how the mechanism works on the issuers, custodians and market makers page, and the underlying plumbing in how Bitcoin ETFs work.

Fee waivers and the small print that ends them

When these funds launched in January 2024 they fought a price war. Several charged nothing at all for an introductory period, and comparison tables written at the time are still circulating with those numbers in them. Every one of those waivers has now expired. If you chose a fund on the strength of a waiver, the single most valuable thing you can do today is open the current prospectus and check what you are actually paying.

VanEck's HODL is the cautionary tale. Its waiver was genuinely aggressive: zero sponsor fee on the first $2.5bn of assets. But it was written to expire on 31 July 2026 whether or not the fund reached that threshold, and it did not come close. HODL held roughly $1.1bn on 30 July 2026 — about 43% of the number in the offer. Holders woke up on 1 August paying the full 0.20%. Nothing went wrong and nobody was misled; the terms did exactly what they said. They were simply written as a date first and a threshold second, and most people read it the other way round.

The others followed the same pattern with less drama. IBIT ran a 0.12% introductory rate on its first $5bn for twelve months from launch; it expired in January 2025, and BlackRock's most recent 10-Q confirms no fees were waived during the first half of 2026. FBTC charged nothing until 31 July 2024. BTCO waived its fee on the first $5bn for six months and that lapsed in July 2024. BITB ran a zero-fee period on its first $1bn. All gone.

The lesson generalises beyond this category. A waiver has three moving parts — a rate, a threshold and a date — and the marketing usually leads with the first, mentions the second and buries the third. Read them in reverse order.

The GBTC problem, and whether to switch

Grayscale's GBTC charges 1.50%. That is six to ten times what its competitors charge for the same asset in comparable custody. On a $100,000 position it is $1,500 a year against $150 at the Mini Trust. There is no performance argument, no structural argument and no service argument that closes a gap that size.

The market has responded accordingly. Since converting to an ETF in January 2024, GBTC has seen roughly $17.5bn of cumulative outflows and has shed something like 447,000 bitcoin. It held 130,481 bitcoin on 1 September 2026. Grayscale's own answer was to spin off about a tenth of the trust's holdings into the Bitcoin Mini Trust at 0.15% and distribute the shares pro rata to GBTC holders — an unusually candid admission that the headline fund was mispriced.

So why does anyone still hold GBTC? Because for a large share of the remaining holders, leaving is expensive in a way the fee comparison does not capture.

The switching decision in a taxable account

GBTC existed as a private trust from 2013 and traded on OTCQX from 2015. Holders who bought then are sitting on very large embedded gains. Selling to move into a cheaper fund realises those gains now. Staying costs 1.35 percentage points a year forever. The break-even depends entirely on the size of the embedded gain relative to the position, and there is no general answer — but the shape of the calculation is simple enough to do on paper.

Take a $50,000 GBTC position with a $20,000 embedded long-term gain. At a 15% federal long-term rate that is $3,000 of tax to switch, ignoring state tax and the net investment income tax. Staying costs the fee difference: 1.35% of $50,000, or $675 in the first year and more as the position grows. Purely on those two numbers, the fee drag overtakes the tax bill somewhere around the fifth year.

Two adjustments matter and both cut against switching too eagerly. First, you reinvest $47,000 rather than $50,000, so you also forgo the return on the tax you paid — the break-even is later than the raw division suggests. Second, and more important: for most people the tax on switching is deferral, not avoidance. You will pay it eventually anyway. The exception is a holder who intends to hold until death, where heirs receive a stepped-up basis, or who plans to donate appreciated shares to charity. For those two cases the tax is genuinely avoided, and the calculation changes completely. If either describes you, the 1.35% is buying something real. If neither does, you are paying an annual fee to defer a bill you will settle regardless.

One caveat on selling at a loss rather than a gain. If you sell one bitcoin ETF at a loss and buy another within thirty days, whether the two are "substantially identical" under section 1091 is genuinely unsettled — two grantor trusts holding the same commodity are an obvious test case that has not been answered. That is a question for your own tax adviser, not for a directory page.

What a fee costs over a decade

The table below holds a $50,000 balance flat for ten years and applies nothing but the sponsor fee. It is not a projection. It assumes bitcoin goes precisely nowhere, which it will not, and it ignores every other cost on this page. Its only purpose is to isolate fee drag so you can see its shape.

The gap between the cheapest fund and the most common fee — eleven basis points — costs about $540 over a decade on this notional. That is real money and it is worth a few minutes of comparison, but it is not life-changing. The gap between the cheapest fund and GBTC is roughly $6,318, or more than an eighth of the entire starting position, surrendered for no benefit whatsoever. That is the number that should move people, and it is why the switching arithmetic above deserves an afternoon rather than a shrug.

Under a rising bitcoin price the absolute drag is larger, because the fee is a percentage of a growing base. Under a falling price it is smaller. The percentages hold either way, which is why comparing funds on percentages rather than dollars is the right habit — and why we have put the dollars in anyway, because percentages do not feel like anything.

The fee is paid in bitcoin, and that has a tax consequence

Here is a detail that almost never appears in fee comparisons and genuinely surprises people. These trusts do not hold cash. To pay the sponsor fee, the trust sells or delivers a small amount of bitcoin — the fee comes out of the asset itself.

Because a spot Bitcoin ETF is a grantor trust rather than a fund registered under the Investment Company Act of 1940, its transactions pass through to shareholders. You are treated as owning your slice of the bitcoin directly. So each of those fee payments is, in proportion to your holding, your disposal of bitcoin — a taxable event. Grayscale's own filings say so in plain terms: each delivery or sale of bitcoin to pay the sponsor's fee is a taxable event for shareholders.

In practice the amounts are small. On a 0.25% fund the annual disposal is a quarter of a percent of your position, and the gain on it is a fraction of that. But the point stands: you can owe a small amount of tax in a year in which you did not place a single trade, and if you were not expecting a line item on your broker statement, it will look like an error. It is not. The mechanics, the reporting and how it interacts with Form 1099-DA are covered in the Bitcoin ETF tax guide. It is also one of the quieter arguments in favour of holding these funds inside a retirement account, where the pass-through disposal has no consequence at all.

Which cost should decide it

The honest summary is that for most people the differences are smaller than the attention they attract — with one enormous exception at the top of the fee table.

Beyond cost, the decision runs into questions this page deliberately does not cover: which custodian holds the coins and how concentrated that is, whether you would rather own the asset outright than a wrapper around it, and how to actually place the trade. Those live on the issuers and custodians page, in Bitcoin ETF versus bitcoin, and in how to buy a Bitcoin ETF. If you want our opinionated read on which fund suits which kind of investor rather than the raw numbers, that is best Bitcoin ETFs. And if you are weighing the fund against other crypto exposures entirely, the full crypto ETF list has the rest of the universe.