YZi Labs (@yzilabs): Opinion: DATs 1.0 were pure Soros. What comes next is pure Buffett.

Opinion: DATs 1.0 were pure Soros. What comes next is pure Buffett. Author: Alex Odagiu, Partner of YZi Labs; Interim CEO of BNC The first generation of digital-asset treasuries ran on reflexivity, and reflexivity always ends. The survivors are becoming three different kinds of company, and the asset built for the most complete one is BNB. The honest description of the first wave of digital-asset treasury (DAT) companies is the one nobody will use: a Soros reflexivity trade dressed as corporate strategy. Not an insult, just what worked. Soros's insight was that price and fundamentals aren't separate: a high enough share price changes what a company can do, which feeds back into the price as long as the operator spins the loop. That is the machine Michael Saylor built, a premium let Strategy issue equity above the value of its holdings, the proceeds bought bitcoin, the bitcoin reinforced the story, the story justified the next round. He named it and ran it harder than anyone alive. The imitators, Metaplanet, SharpLink, Semler and a long tail, copied the loop with varying competence. A few built something real; many issued below the value of their holdings and became wrappers around dilution. But the obvious hasn't sunk in: reflexive loops expire. Reflexivity is starting capital, not a business. It ran on four premiums, each thinner in 2026 than in 2022: access (until spot ETFs arrived for bitcoin, ether, now BNB, XRP and SOL), interpretation (until allocator literacy caught up), scarcity (until look-alikes commoditised the wrapper), and cheap financing (until rates rose and the trade crowded). Remove them and a company whose only product is its own premium is in trouble: the loop reverses, and what compounded on the way up de-compounds on the way down, the arc of every reflexive trade, from 1960s conglomerates to the SPAC boom. Strategy and BitMine are not reflexivity vehicles anymore. Both saw the expiry date coming and rebuilt around it, and, tellingly, they did not rebuild into the same thing. Strategy is a digital-credit engine Strategy is now a digital-credit engine on a bitcoin-heavy balance sheet. Bitcoin is a near-pristine reserve asset, no counterparty, no issuer, continuous settlement, cryptographically verifiable — and Saylor saw the financing implications early. The preferred family, STRK, STRF, STRD and above all STRC, is the proof: each reaches a different investor, including mandates that will never own a coin or touch MSTR common. Here the language must be exact, because the shorthand is wrong. STRC is not secured by Strategy's bitcoin: the disclosures say the preferred securities aren't collateralised by bitcoin and carry only a claim on residual assets. Bitcoin provides economic backing, not legal collateral, an unsecured credit stack on an exceptionally hard, if volatile, reserve asset, raised against as long as the market trusts it. And the behaviour now matches the structure: in late May Strategy sold 32 bitcoin to help fund distributions, then a week later, on 8 June, bought 1,550 more and topped its cash reserve to $1 billion. A treasury that sells and buys to manage its book isn't running "HODL" gospel but asset-liability management. Preferred must be serviced, yet it reduces Strategy's dependence on an equity premium, which is why the premium-to-NAV question everyone obsesses over matters less than it did. Saylor has built, in public-equity form, something close to a bank denominated in bitcoin. BitMine is an operating-yield engine BitMine took the other road. Its tell is MAVAN, pitched as institutional staking infrastructure, not a pile of tokens. A reported 5.74 million, with 4.88 million staked, stops being a static reserve and becomes a productive asset that throws off cash flow, the Berkshire move, where operating cash flow and cheap insurance float, not stock-picking, were Buffett's edge. Ask "why not just own ETH?" and the answer is that BitMine owns the machinery producing yield at scale, a business whether ether trades at $4,000 or $40,000. Then this month it reached for Saylor's other tool: 3.5 million 9.50% Series A perpetual preferred shares at $80 (about $273.8 million), $100 stated value, weekly cash dividends, NYSE listing as BMNP. The economics expose the constraint: 9.50% on a $100 amount sold at $80 is a cash cost near 11.9%, while BitMine's reported staking yield runs in the low single digits, in line with the 3–5% the market pays. This is no carry trade; it works only as balance-sheet architecture, non-common capital to grow ether exposure and fund infrastructure, hoping to lift ether per share after cost. Operating yield is real, but it doesn't make the cost of capital vanish. Both point the same way: STRC and BMNP open a DAT credit curve, each layer a claim that must be serviced and judged on whether it raises digital assets per share. DAT 2.0 isn't the end of reflexivity so much as its institutionalisation, the naïve premium loop replaced by a structured balance sheet. The third form, and the asset that needs it Neither a credit engine nor an operating-yield engine is the complete version. That is Berkshire-shaped, one balance sheet running all three of Buffett's engines at once: float, operations and allocation. In Buffett's hands, float is cheap, long-dated liability, the insurance premiums he invests before claims fall due, operations are the cash-generating businesses, and allocation is the deployment of both. In crypto: digital float is the credit stack of perpetual preferreds and converts; digital operations are the validators, staking, custody, market-making; and digital allocation is deployment across the ecosystem, all of it run on a reserve-asset treasury, not on the treasury itself standing in for float. Strategy has the float and the reserve but no yield-generating operations; BitMine has operations and a reserve, and has just opened its own float with BMNP, but isn't yet a multi-business allocator. Nobody has built the synthesis, the form least exposed to any single premium compressing, because it compounds through more than one engine. Which asset is it built for? Bitcoin fits the credit engine: the asset is the product. Ether fits the operating company: staking is its dominant productive activity. BNB is the unusual one. It sits at the centre of a broad cash-flow surface: exchange utility, on-chain activity, one of the largest share of stablecoin transactions, DeFi, asset issuance, agentic finance. It adds a yield surface wider than bitcoin's, active deflation (quarterly burns toward 100 million tokens), US spot listing. 1) Collateral utility, 2) yield, 3) ecosystem cash flow, 4) deflation: bitcoin has the first, ether the second, but BNB is the rare asset that carries meaningful amounts of all four at once. A BNB treasury that stops at a credit engine or an operating company leaves layers unbuilt; BNB requires the full Berkshire structure to express. A Berkshire-shaped BNB company would turn balance-sheet BNB into cash flow analysable on ordinary terms, allocate across the ecosystem, and treat BNB as a productive reserve, compounded by operations, allocation and the burn. The BitMine preferred should function as a warning rather than a template. A BNB issuer cannot wave at its "broad yield surface" and imply the dividends fund themselves. The monthly test is one question, does each issuance, stake or dividend raise long-term BNB per share, or just the token count? A BNB credit stack could be genuinely powerful, it could open the asset to income investors who would not touch it directly, but only if it is built as a transparent credit product and not as a promotional extension of the premium-to-NAV trade. Strategy proved a bitcoin balance sheet can become a credit engine, and that once it does, you stop being a believer and start being an asset-liability manager. BitMine is proving an ether treasury can become an operating-yield engine. The third model is unbuilt, and BNB is the most plausible asset to build it around. The winner won't be whoever owns the most tokens or issues the most paper; it will be whoever can show, month after month, that the whole capital structure is still raising digital assets per share. The easy Soros trade is over. The Buffett digital balance sheet is what comes next.