Spark bets on stablecoin fragmentation

Spark, the DeFi lending arm tied to Sky (formerly MakerDAO), has quietly shelved its consumer app and rebuilt itself as backend plumbing for companies that already own the customers. Instead of competing with Coinbase, PayPal and Robinhood for users, it now supplies them with yield and liquidity behind the scenes.

The numbers behind the pivot

Spark's annual revenue fell from about $80 million in the bull market to roughly $20 million now, according to Phoenix Labs CEO Sam MacPherson in an interview with CoinDesk. Against that drop, two lines are growing. Its Bitcoin-backed OTC lending book through Anchorage sits at $260 million outstanding, up from roughly $400 million originated total, with a $1 billion year-end target that implies nearly quadrupling the outstanding balance. Its stablecoin FX layer on Uniswap v4, built around a hook called DualPool, moved about $150 million into pools pairing USDS against USDT and PYUSD and routed roughly $1.5 billion in swap volume in its first 30 days, accounting for close to 30% of stablecoin-to-stablecoin swap volume on Uniswap. Separately, Robinhood's Earn product, which routes USDG deposits through a Morpho vault that includes Spark's spUSDG, has pulled in more than $200 million in deposits in 24 days at roughly 7% APY.

Infrastructure over ownership

The read here is straightforward: Spark decided owning distribution wasn't worth the fight, so it's selling the thing every issuer needs once distribution fragments, which is a way to move money between tokens without users noticing. PayPal, Robinhood, Coinbase and Stripe are each building or backing their own stablecoin rails, and MacPherson expects the field to fragment further as GENIUS Act rules take effect and the Clarity Act potentially advances. That scattering is bad for any single issuer trying to concentrate liquidity, but it's the entire business case for a neutral routing layer sitting underneath all of them. Spark doesn't need to win users. It needs every issuer to eventually need a bridge, and it's positioning to be that bridge before the fragmentation fully plays out. The revenue drop from $80 million to $20 million shows the model isn't free money yet, but the OTC and swap-routing numbers are the parts actually compounding.

What confirms or kills this

The clearest test is whether Spark's OTC lending book actually reaches $1 billion by year-end from its current $260 million outstanding. A shortfall would suggest institutional demand for Spark-brokered credit is thinner than MacPherson's framing implies. If it hits or nears the target while Uniswap swap-routing volume keeps climbing past its first-30-days pace, that's evidence the backend bet is working faster than the consumer-app path ever did.