Intesa Sanpaolo Cut IBIT 94%, Tripled Ether ETF in Q2

Intesa Sanpaolo's second-quarter holdings disclosure carried two lines that, read together, tell a different story than the wire copy has been running. The first: a 94% reduction in the bank's position in BlackRock's iShares Bitcoin Trust. The second, further in the same document: the Ether ETF stake, tripled. Both moves landed in a quarter when spot crypto was already sliding. I spent an afternoon reading that filing next to the venue-level fee schedules where the underlying assets actually settle, and the pattern is not the one the headlines pushed. This is a rotation. And the rotation follows the fee curve, not the price.

Methodology

What I did was narrow. I took the two headline numbers from the Q2 disclosure — the 94% IBIT cut and the tripled Ether ETF position — and I put them next to the fee schedules and license footprints of the five venues where the underlying spot flow actually clears. Binance, Bybit, Bitget, OKX, MEXC. That is the reference frame I trust, because those are the venues whose maker/taker curves define the arbitrage economics that authorized participants use when they create or redeem shares of the ETF wrappers Intesa was trading.

I did not model Intesa's P&L. I did not try to back out an entry price. I did not read internal treasury minutes because I do not have them and nobody in the wire copy has them either. What I did instead is a much less glamorous exercise — I asked whether the venue-level economics of BTC versus ETH could account for the direction of the rotation, independent of price. The answer is more interesting than the headline framing.

Limitations up front: the underlying licensing data I use here is current as of the last venue audits I could pull, and the fee schedules are the public tiered rates before market-maker rebates. Intesa's actual execution costs are almost certainly better than retail. That does not change the direction of the argument. It changes the size of the effect.

Finding #1: The 94% Cut Is Not a Retreat — the Basket Weight Moved, Not the Basket

Ninety-four percent is the kind of number that reads like a slammed door. The story the wires wrote around it was the obvious one — bank pulls back, risk-off, crypto winter reflex. That reading requires you to look at the IBIT line alone and stop reading before you get to the Ether page of the same document.

The Ether ETF position tripled in the same quarter. Both moves are in the same filing. Both are Q2 activity. If this were a retreat from crypto as an asset class, the Ether leg would have been trimmed, held flat, or quietly liquidated in the same window. It was not. It was scaled up.

What this actually looks like is a basket-weight decision. Somebody at Intesa's treasury or asset-management arm decided that the ratio between the two exposures needed to change, not that the aggregate exposure needed to go to zero. That is a fundamentally different signal — and it is the signal the ranking sites and the "institutions are dumping crypto" takes are burying under headline arithmetic.

The question that follows is: why this direction. Why cut BTC exposure 94% and triple ETH exposure, in a quarter where BTC's price was falling faster than ETH's on some intraday windows? The pure directional-bet answer does not explain the shape of the rotation. If you thought BTC was going lower, you cut BTC — you do not simultaneously add three times the ETH exposure unless the underlying thesis is about something other than spot direction. The rest of this piece is an argument that the something-else is fee-curve and custody economics, not price.

Finding #2: The Ether Leg Was Bought Into a Price Slump, Which Rules Out One Explanation

Here is the part that eliminates the lazy read. The Ether stake tripled during a quarter when crypto prices were sliding — not before, not after. This rules out the momentum explanation. It rules out the "we saw a rally coming" explanation. Nobody triples an ETF position into a falling tape because they think the tape is about to reverse — that is not how bank treasury desks size risk.

What does happen during a slump: the fee-per-dollar-of-turnover economics of the underlying venues start mattering more, because volume-driven revenue is compressing and every basis point of execution cost is proportionally larger. If you are running an ETF wrapper strategy where the P&L is a spread over the underlying venue costs, the venues where the underlying trades — Binance, OKX, Bybit — become the actual determinant of whether the wrapper is worth carrying.

Now look at what the venue fee curves say about BTC-heavy versus ETH-heavy positioning. Binance and Bybit both charge 0.10% maker and 0.10% taker at the base tier. OKX is 0.08% maker and 0.10% taker. Bitget matches Binance at 0.10%/0.10%. MEXC undercuts everyone with 0.00% maker and 0.02% taker. These are the schedules against which any authorized participant creating or redeeming crypto ETF shares is running their arbitrage math.

I could not pull the current tiered rates that a bank of Intesa's size would actually pay — those are negotiated privately, and the public schedules understate the discount by an unknown amount. But the ratio between venues is preserved down the tiers. And that ratio matters more for ETH than for BTC, because ETH's on-venue liquidity fragmentation is different — a fact I am about to unpack in the fee-curve math.

Finding #3: Where the Underlying Actually Settles — the Venue Fee Curve Behind the Rotation

This is the math teardown. Follow every number.

Start with the daily volume figures. Binance clears $18.5 billion per day. Bybit clears $9.2 billion. OKX clears $4.9 billion. Bitget clears $6.1 billion. MEXC clears $3.8 billion. Aggregate across the five: $42.5 billion per day of self-reported spot and derivatives turnover.

Now the fee-weighted view. If an authorized participant is executing $100 million of BTC creates through the tier-1 venues at the public taker rate, the cost is: $100M × 0.10% = $100,000 at Binance, $100,000 at Bybit, $100,000 at Bitget, $100,000 at OKX, $20,000 at MEXC. Five-venue average, equal-weighted: $84,000 per $100M of turnover, or 8.4 basis points.

If the same $100M is executed as maker liquidity at Binance's public rate — $100,000, because Binance's maker and taker are both 0.10% — versus Bitget's $100,000 versus OKX's $80,000 versus MEXC's $0. The maker-side blended cost drops to 5.6 basis points on the same equal-weighted mix. The gap between taker and maker execution across this venue set is 2.8 basis points per turn.

That 2.8 bp gap compounds. An ETF wrapper that rebalances the underlying weekly, at the maker-side blended rate versus the taker-side, saves 2.8 × 52 = 145.6 basis points per year of gross turnover cost. Not on the ETF's NAV — on the turnover it is doing to maintain the NAV. That is the number that lives on the wrapper operator's income statement, and it is the number that shifts when you change which underlying you are wrapping.

Here is why it matters for the BTC-to-ETH rotation specifically. ETH's higher venue-count fragmentation — 620 pairs on Bybit alone, 720 on OKX, 2,400 on MEXC — means that a wrapper carrying ETH exposure has more venues to source liquidity from at the maker side. BTC's liquidity is concentrated in fewer, deeper books. That sounds like a BTC advantage, and for pure execution it is. For fee-arbitrage economics on a wrapper vehicle, it is the opposite — concentrated books mean less maker-side rebate opportunity, because the queue is shorter and rebates get eaten faster.

The rotation Intesa executed — cutting the concentrated-book exposure 94%, tripling the fragmented-book exposure — is the rotation you would expect if the underlying decision was optimizing wrapper economics rather than expressing a directional view.

Finding #4: The Custody Incentive Nobody Is Naming in the Wire Copy

Now the incentive layer. This is where I have to be careful, because the moment you say "custody" the retail-Twitter crowd assumes you mean self-custody versus exchange custody. That is not the question in this filing. The question is qualified custodian economics.

Coinbase Custody is a New York DFS Trust Company. Fidelity Digital Assets holds the same charter. Anchorage Digital holds an OCC Federal Trust Charter — the first crypto bank. These three are where a bank of Intesa's size can legally park spot exposure without carrying the operational burden of custody themselves. Every crypto ETF wrapper of size uses one of these three, or a similar qualified custodian.

Qualified custodians charge fees. Those fees are structured differently for BTC than for ETH — not because of the assets, but because of the operational tail. BTC custody is dominated by cold-storage multisig operations whose marginal cost per unit of AUM is close to zero above a certain threshold. ETH custody involves smart-contract interaction, staking operations, and validator management, which the custodian can monetize as an additional revenue line and pass through as a lower base fee. The net effect: for an ETF wrapper operator, ETH exposure often carries a lower all-in custody cost per dollar of AUM than BTC exposure, once staking revenue is netted against custody fees.

I did not pull Intesa's specific custody agreements — those are not public. But the direction of the incentive is public, structural, and stable. When a bank rotates from a BTC wrapper to an ETH wrapper in a slumping market, the plausible unstated reason is that the ETH wrapper's total-cost-of-carry is lower after custody-side offsets. The wire copy that frames this as "risk-off" or "sentiment shift" is skipping the layer where the actual money is being made and lost.

Follow the incentives. Nobody at a European systemically important bank triples an ETH ETF position in a falling market because they got a hunch. They do it because a spreadsheet somewhere showed the net-of-fees carry looked better than the alternative. The 94% IBIT cut is the other side of that same spreadsheet.

What This Does NOT Prove

I want to draw a hard line around what this argument does and does not establish. It does not prove that Intesa Sanpaolo's treasury committee actually reasoned in fee-curve terms when they made the rotation. I have no minutes. I have no phone calls. I have a filing and a set of venue schedules and a plausibility argument that connects them. Plausibility is not testimony.

It also does not prove that the rotation was a good decision. The wrapper-economics argument works if the venues' fee structures hold, if the custody-side offsets behave as described, and if the rotation is held long enough for the basis-point savings to compound past whatever transaction cost was paid to execute the rotation itself. Any of those assumptions can break. If ETH's on-venue fragmentation reverses — if the market consolidates back onto Binance and OKX at the expense of MEXC — the wrapper-economics case gets weaker.

And the 94% figure itself carries a rounding tail I could not verify against a share-count breakdown, only against the percentage disclosure. If the actual reduction is somewhere between 91% and 96%, the analysis does not change. If it is somehow materially outside that range, someone with access to the underlying custody statement should say so.

The Takeaway

The rotation is real, the direction is asymmetric, and the wire-copy framing is missing the layer where the decision actually got made. Follow the fee curve.

FAQ

Why does the 94% IBIT cut matter if the Ether position tripled in the same filing?

Because the two numbers together rule out the "bank pulled back from crypto" reading that most coverage ran with. A retreat from the asset class would show both positions trimmed or held flat. A basket-weight rotation shows one leg cut hard and another leg scaled up. The direction of the rotation — out of BTC exposure, into ETH exposure — is the signal worth reading, not the headline percentage on either line alone.

Does the venue fee data in this article apply to institutional execution?

Partially. The maker/taker rates cited — 0.10%/0.10% on Binance, Bybit and Bitget, 0.08%/0.10% on OKX, 0.00%/0.02% on MEXC — are the public tier-one schedules. Institutions of Intesa's size negotiate discounted rates that are not public. The ratios between venues generally hold down the tiers, so the direction of the argument survives. The absolute basis-point figures do not.

Why is ETH custody cheaper than BTC custody for a wrapper operator?

Because qualified custodians like Coinbase Custody, Fidelity Digital Assets, and Anchorage Digital can monetize ETH holdings through staking and validator operations, and pass part of that revenue back as a lower base custody fee. BTC has no equivalent yield-generating primitive at the custodian layer. The net all-in cost of carrying ETH exposure at a qualified custodian can be lower than BTC, once staking offsets are netted — which flips the intuition most retail readers start with.

Does this analysis prove Intesa made the right call?

No. The article argues that the rotation is consistent with a fee-curve and custody-economics reasoning, not that the reasoning was correct or that the outcome will be favorable. The case depends on venue fee structures holding, custody offsets behaving as expected, and the rotation being held long enough for the basis-point savings to compound past execution costs. Any of those can move against the position. Plausibility of motive is not the same as validation of outcome.

Whether the wrapper-economics thesis actually explains the Q2 rotation — or whether a treasury memo somewhere describes a completely different reason nobody outside Intesa has seen — is a question the public filing does not settle. If you have seen that memo, write.