Perp DEXs on Solana in 2026: Jupiter, Drift and Pacifica

Solana’s quiet bid to out-trade the field

Most of the attention in on-chain derivatives flows to Hyperliquid, and the price action explains why. Its HYPE token trades near an all-time high around $96.90, for a market value north of $21 billion, according to CoinGecko. But a quieter contest has been running one chain over. Solana, the network built for cheap and fast blockspace, has become the second-largest home for perpetual futures traded on-chain, behind only Hyperliquid, and it got there without any single venue grabbing the spotlight.

The scale is real. By late August 2026, Solana-based perpetual platforms had cleared more than $1.08 trillion in cumulative notional volume, a figure built across several venues rather than in one spike, CryptoBriefing reported. SOL itself changes hands near $118.94, a top-ten asset by market value on CoinGecko. The leveraged-trading economy sitting on top of it is now large enough to move the whole network’s fee revenue and to draw serious builders.

It has also been tested, hard. In April 2026, Drift Protocol, one of Solana’s flagship perp venues, lost roughly $285 million in the biggest DeFi hack of the year. This guide maps the Solana perp landscape as it stands in the autumn of 2026: the two designs that split the market, the three venues that dominate it (Jupiter, Drift and Pacifica), what the Drift breach revealed about the real risks, how to read the volume numbers without being fooled, and where the US regulator, the CFTC rather than the SEC, draws its line.

What a perp DEX is, in one minute

A perpetual future is a derivative that tracks an asset’s price with leverage and no expiry date. Instead of settling on a fixed date like a traditional future, it relies on a funding rate, a small periodic payment between longs and shorts, to keep its price tethered to the underlying spot market. When the perp trades above spot, longs pay shorts; when it trades below, shorts pay longs. That single mechanism is what lets a contract run indefinitely without drifting away from the thing it is meant to price.

A perpetual DEX offers those contracts through smart contracts instead of a company. There is no central operator holding your balance, no account to be frozen, and in most cases no identity check at the door. You connect a wallet, post margin (usually a stablecoin like USDC), and open a position that the protocol tracks on-chain against a mark price. If losses eat through your margin, the protocol liquidates you automatically, and how much cushion you keep depends on whether you trade with isolated margin (risking only that position) or cross margin (backing it with your whole balance). The appeal is self-custody and open access; the cost is that when something breaks, there is rarely anyone to call. The rest of this piece assumes those basics and focuses on how Solana’s venues put them into practice.

Liquidation is the mechanism that keeps the whole system solvent, and it is worth understanding before you touch leverage. Every position has a liquidation price, the level at which your margin no longer covers your losses. Cross that line and the protocol closes your position, often at a worse price than you would choose, and takes a penalty on top. The higher your leverage, the closer that line sits to the current price, so a position at 50x can be wiped out by a move of a couple of percent. On a volatile day, a cluster of liquidations can feed on itself, pushing the price further and triggering the next wave, which is why perp venues cap leverage and lean on insurance funds or backstop pools to absorb the overflow.

Why Solana, and not a dedicated appchain

Hyperliquid answered the perp-DEX design problem by building its own blockchain, a purpose-made Layer 1 with a matching engine baked into consensus, so its order book runs fully on-chain at sub-second speed. Solana’s builders made the opposite bet. Rather than launch a dedicated appchain, they put their exchanges on a shared, general-purpose network that already produces blocks in roughly 400 milliseconds and charges fractions of a cent per transaction, with Pyth price feeds available natively.

Solana has been chasing this use case for years. Its earliest flagship was an order-book exchange, and the tooling that grew up around it, fast matching, cheap cancels, and a culture of professional market making, left the network unusually well suited to derivatives. Traders who wanted the feel of a centralized exchange, quick fills and tight spreads, without handing over custody, found Solana came closest. That heritage is a big part of why the network, rather than a general-purpose Layer 2, emerged as the main on-chain rival to a purpose-built chain like Hyperliquid.

That choice carries trade-offs in both directions. On the plus side, a Solana perp venue inherits the chain’s throughput, its native oracles, and deep composability: the liquidity-provider token that backs one exchange can be posted as collateral on another, and capital moves freely between lending, staking and trading. On the minus side, these venues share blockspace with everything else on Solana, so they are exposed to congestion, priority-fee auctions during busy moments, and the network’s history of outages in a way a dedicated chain is not. When a memecoin frenzy clogs the network, a perp trader’s liquidation can land in the same queue as thousands of unrelated transactions.

The custody model is worth holding onto as well. On a Solana perp DEX you sign every order from your own wallet, the same self-custody approach that smart-account wallets are trying to make safer and less error-prone. Composability cuts both ways, though. When Drift was drained in April, some of the stolen assets were Jupiter’s own pool token, a reminder that on a shared chain, one venue’s failure can reach straight into another’s balance sheet.

Two designs on one chain: pools versus order books

Strip away the branding and Solana’s perp venues fall into two families, separated by who takes the other side of your trade.

The first is peer-to-pool. There is no order book. You trade against a shared pool of liquidity at a price fed in by an oracle, and the people who deposited into that pool are your counterparty. Jupiter’s exchange is the flagship example, and the model descends directly from the GMX design that popularized it on other chains. The second is the order book, whether fully on-chain or hybrid. Buyers and sellers post bids and offers, professional market makers provide most of the depth, and the exchange matches them. Drift runs an on-chain order book supplemented by a just-in-time auction and an automated market maker as a fallback; Pacifica runs a fast off-chain matching engine that settles finished trades back to Solana.

Each design decides three things differently: how orders are matched, where the price comes from, and who absorbs the risk when traders win. Those choices shape everything downstream, from execution quality to the failure modes that show up in a crisis. A hybrid model like Pacifica’s buys centralized-exchange speed at the cost of a matching layer that traders have to trust to sequence orders fairly. The table below lays out the three venues that dominate Solana perps today.

Jupiter Perps and the JLP house

Jupiter is the largest perp venue on Solana by a wide margin. By mid-2026 it was handling roughly 80% of the network’s perpetual-futures volume, according to CryptoBriefing, and its perpetual exchange has grown into one of the busiest in all of DeFi, as tracked on DeFiLlama. It rose to that position on the back of Jupiter’s dominance as Solana’s main trading aggregator, which funnels order flow straight into the perp product.

The engine is the JLP pool, a basket of SOL, ETH, wrapped BTC and stablecoins that at times has held well over $1.5 billion in deposits. When you open a leveraged long on Jupiter, you are not matched against another trader; you borrow exposure from JLP and trade at the oracle price with no order-book slippage. That makes execution clean, even in size, because the pool absorbs the position rather than walking up a thin book. The flip side is that JLP is the house. When traders as a group win, the pool pays them; when traders lose, and over time most leveraged traders do, those losses flow to JLP holders, on top of the trading and borrowing fees that make up the bulk of the pool’s income. Roughly three-quarters of platform fees are routed to LPs.

The asymmetry is worth sitting with, because it defines who wins over a full cycle. Imagine a hundred traders open leveraged longs into a rally; the pool pays them out, and JLP holders take the loss even as fees roll in. Now imagine that same crowd gets liquidated in a sharp reversal; their margin flows to the pool, and JLP has a very good week. Over long stretches the fees and the net trader losses have tended to outweigh the payouts, which is why JLP has been a viable yield product, but the return is lumpy and it correlates with the pain of the traders on the other side. Depositing into JLP is a bet that the house edge holds.

That structure is why JLP is often pitched as a source of real yield: its returns come from fees and net trader losses, actual cash flows, rather than from a protocol minting its own token to pay depositors. The distinction matters, and we have unpacked it before in the context of real yield after the Fed hike. JLP has paid double-digit annualized returns in good stretches, which is what drew deposits in the first place. But it is not a savings account. Half the basket is volatile crypto, so holders carry directional exposure, and the pool is only as safe as the oracles that price the positions traded against it. A sustained run of winning traders, or one bad price print, lands squarely on the pool.

Drift Protocol’s order-book bet

Drift took the harder engineering path: a real order book on Solana. Rather than a single pool, it stitches together three sources of liquidity. A central limit order book lets market makers post resting bids and offers. A just-in-time auction invites market makers to compete to fill each incoming order at the best available price within a short window. And an automated market maker sits behind both as a backstop, so there is always a quote even when human liquidity thins out. Traders can push leverage as high as 101x, per CryptoBriefing.

The three-layer design is Drift’s answer to a hard problem: order books need liquidity to feel liquid, and a young exchange rarely has enough market makers to keep one deep. The just-in-time auction lets professional firms step in only at the moment of a trade rather than posting capital around the clock, while the automated market maker guarantees a fallback price so no order goes unfilled. Combined with cross-margining and support for spot as well as perps, that let Drift feel like a full exchange rather than a single product, and it is a big reason the protocol accumulated the deposits and integrations it did.

Bad debt, the risk that a liquidation cannot be closed at a good enough price, is absorbed by an insurance fund that users stake into for a share of protocol revenue. The DRIFT token coordinates governance and incentives. For most of 2025 and early 2026, Drift was one of the two engines of Solana derivatives volume alongside Jupiter, and it was woven deeply into the rest of Solana DeFi, running delta-neutral and super-staking vaults that recycled assets from across the ecosystem into yield strategies. That integration made it powerful and capital-efficient. It also meant that when the protocol was compromised, the damage did not stay contained to a single balance sheet.

The $285 million lesson: how Drift lost its keys

On April 1, 2026, Drift lost roughly $285 million, more than half of its total value locked, in what became the largest DeFi hack of the year. The striking part is that it was not a smart-contract bug. The code did exactly what it was told; the problem was who got to give the orders.

According to a post-incident analysis by Chainalysis, the attackers spent months posing as a trading firm and building relationships with Drift contributors, then abused a Solana feature called durable nonces to get members of the protocol’s security council to unknowingly pre-sign transactions. Durable nonces let a transaction be signed now and submitted later, which is handy for legitimate coordination but devastating when the thing being signed hands over administrative control. With admin rights in hand, the attackers turned to a token they had created weeks earlier, on March 12, named CarbonVote (CVT). They controlled about 80% of its supply, wash-traded it to a fake price of $1, and controlled the oracle reporting that value. They then whitelisted CVT as collateral with extreme borrowing limits, deposited 500 million worthless tokens, and used them to withdraw around $285 million in genuine assets, including USDC, wrapped BTC and Jupiter’s JLP, drained from three vaults before the funds were swapped to USDC and bridged to Ethereum.

Blockchain-forensics firm Elliptic put the loss at $286 million and tied the operation, with what Drift itself called medium-high confidence, to the North Korean state-linked group behind the October 2024 Radiant Capital hack, tracked as UNC4736. Elliptic called it the eighteenth DPRK-linked theft it had recorded in 2026. In other words, this was a keys-not-code failure, the same category as the year’s other headline losses, including the $320 million Liquid Network bridge bug. On a shared chain the blast radius was wider: because JLP was among the stolen assets, Jupiter’s liquidity providers were pulled into a breach that began somewhere else entirely.

The method matters as much as the money. Through 2026, the most expensive crypto losses have come not from clever math against smart contracts but from people: operators tricked, keys phished, and insiders planted or impersonated. North Korean crews in particular have industrialized the social-engineering playbook, spending months embedded in a project’s community before making a move, exactly the pattern that played out at Drift. For a Solana trader the lesson is uncomfortable but simple: the safety of a venue depends less on how audited its code is and more on how its administrative keys are held, rotated and monitored.

The recovery, and who actually pays

What happened next is a live case study in whether a hacked DeFi protocol can make its users whole. Rather than reimburse everyone at once, Drift laid out a staged recovery. Stablecoin issuer Tether committed up to $127.5 million and other partners a further $20 million toward the shortfall, according to BeInCrypto, with affected users receiving claim tokens to be redeemed over time from protocol revenue and outside funding. The protocol said its separate insurance fund was untouched and that depositors there would keep their assets when trading resumed.

Not everyone was satisfied. Spreading repayment across future revenue means some users wait, and how long depends on how quickly the relaunched exchange earns. Cindy Leow, Drift’s co-founder, framed the rescue as a vote of confidence, saying the willingness of partners “to commit real capital to Drift’s recovery says something about the strength of what we’ve built and what we’re building next.” Whether that confidence is repaid, literally, is one of the more important open questions in Solana DeFi heading into 2027, and a template other protocols will study the next time keys go missing.

Pacifica and the no-token challengers

If Jupiter is the incumbent and Drift the wounded veteran, Pacifica is the insurgent. Launched in 2025 and built natively on Solana, it stands out for what it does not have: no venture-capital backers, no investor token allocations, and, so far, no token at all. According to KuCoin’s project profile, the team includes operators who previously worked at FTX, Binance, Coinbase, Jane Street and Fidelity, and the exchange leans on AI-assisted trading tools rather than a governance coin to set itself apart.

That combination made it a magnet for traders. Pacifica grew fast enough to overtake Jupiter in daily volume at points, cleared more than $100 billion in cumulative volume by early 2026, and reached roughly $18 billion in 30-day volume, ranking around ninth among all perpetual DEXs. Yet its total value locked sits under $30 million, a gap that looks alarming until you remember it runs an order book, not a pool, so it does not need a large deposit base to function; market makers, not depositors, supply the liquidity.

Pacifica is not alone in that model. A whole cohort of perp venues has launched in the past two years running a points program in place of a live token, on Solana and beyond, on the theory that a generous future airdrop is the cheapest way to buy early liquidity and attention. It works, up to a point: volume and users arrive quickly. The question every one of them faces is what happens the day the token lands and the incentive to churn disappears. For Pacifica, with real trading tools and a credentialed team, the bet is that enough of the activity is genuine to survive that transition.

The distance between enormous volume and thin locked value points straight at the catch. With no token yet, much of Pacifica’s activity is points farming, traders churning volume to qualify for a future airdrop. The AI tooling, automated strategies and analytics built into the interface, is aimed at active traders rather than passive depositors, which fits a venue with no coin to market. None of that is unique to Pacifica, and the churn-for-airdrop pattern is exactly what makes raw volume the least trustworthy number in this sector. Which is where reading the scoreboard comes in.

Reading the scoreboard: volume, open interest, TVL

Volume is the headline number every perp DEX wants you to see, and it is the easiest one to inflate. Wash trading, where the same entity buys and sells to itself, points programs that reward churn, and self-reported figures that no outsider can audit all pump the top line without a single genuine risk being taken. The clearest recent lesson came from off Solana: in October 2025, DeFiLlama removed Aster’s perpetual volumes from its rankings over suspected wash trading. DeFiLlama’s founder, known as 0xngmi, noted that Aster’s volume in some pairs tracked a large centralized exchange almost one-to-one, while Hyperliquid showed a “more decorrelated profile,” and that Aster had supplied no lower-level order data to verify the figures, as Bitcoin.com reported.

The fix is to weigh metrics that are harder to fake. Open interest, the total value of positions currently open, reflects real capital at risk and cannot be churned away in a loop; a venue with huge volume but small open interest is usually being farmed rather than genuinely used. Protocol revenue and liquidity-pool profit and loss show whether an exchange is actually earning. Total value locked matters for pool-based venues like Jupiter but says little about an order-book venue like Pacifica. The table below is a quick reference for what each number really tells you.

A single ratio does a lot of the work here: volume divided by open interest. A venue where roughly the same positions turn over a few times a day, a ratio in the low single digits, looks like genuine trading. A venue posting tens of dollars of volume for every dollar of open interest is almost certainly being farmed, because real traders do not flip their entire book dozens of times a session. When Pacifica’s headline volume towered over its open positions during airdrop season, that gap was the tell, the same signal that flagged Aster off Solana a year earlier.

What a Solana perp trade costs, after the Fed hike

Trading a Solana perp is cheaper than most centralized alternatives, but it is not free, and the true cost has several parts. There are trading fees (a maker or taker fee on order-book venues, or borrowing and price-impact fees on Jupiter’s pool). There is the spread and any slippage. There are Solana’s priority fees, which climb when the network is busy, so a volatile day can quietly raise your execution cost. And there is the oracle: because these venues price your position from an external feed, a lag between the oracle and the real market can cost you at the exact moment of a liquidation, and in the worst case a manipulated feed is the whole attack, as the fake CVT price showed at Drift.

The largest variable, though, is funding. Funding rates move with market positioning, but they also sit on top of the prevailing risk-free rate, and that floor rose in 2026. The Federal Reserve raised its policy rate to a range of 3.75% to 4% in September, its first hike since 2023, per the Fed’s own statement. When cash earns more, the cost of carrying a leveraged long relative to simply holding Treasury bills goes up, and funding on crowded long positions tends to follow. Higher rates make cheap leverage a little less cheap, a headwind that applies to every perp venue, on Solana or off.

It helps to see funding as rent on borrowed exposure. If a perp trades persistently above spot because longs are crowded, those longs pay a stream of small payments to shorts, sometimes several times a day, and over a held position that rent compounds. In a calm, bullish market the annualized cost can run into the double digits, which quietly eats returns even when the trade is right on direction. With Treasury bills now paying a real yield, the hurdle a leveraged long has to clear just to break even is higher than it was during the zero-rate years, and disciplined traders price that in before they size up.

One cost the offshore era often ignored is now harder to dodge: tax. A perpetual position that closes at a profit is a taxable event, and US traders face reporting obligations that the no-identity design of these venues does nothing to erase, as our guide to US crypto tax in 2026 lays out. The protocol will not send you a form, but the liability is yours all the same.

Where the US regulator draws the line

The jurisdictional question trips up a lot of people, so it is worth stating plainly. In the United States, perpetual futures are derivatives, which puts them under the Commodity Futures Trading Commission, not the Securities and Exchange Commission. The SEC’s interest, where it has one, is in whether a venue’s token is an unregistered security, not in the contracts themselves. On the product, the CFTC is the relevant regulator.

And the CFTC has started letting perps come onshore. In May 2026 it approved a regulated, exchange-listed Bitcoin perpetual from Kalshi, the first of its kind on a registered US venue. Michael Selig, the agency’s chairman, described the approval as a “watershed moment” and noted that the CFTC had not cleared a new type of derivative in over a decade, according to CoinDesk. The broader market-structure bill that would settle the CFTC and SEC boundary, the CLARITY Act, stalled the same autumn when a Senate procedural vote fell short of the sixty needed to advance it, CNBC reported.

Step back and the direction of travel is clear. On one side sit regulated, identity-checked perpetuals from names like Kalshi and, increasingly, established brokers, cleared by the CFTC and wrapped in the usual investor protections. On the other sit permissionless, anonymous venues like the Solana trio, open to anyone with a wallet and answerable to no one. The two are converging on the same product, the perpetual future, from opposite ends of the compliance spectrum, and the middle ground between them is thin. A trader’s real choice in 2026 is less about which venue has the best fees than about which of those two worlds they want to live in.

None of that reaches the Solana venues in this guide. Jupiter, Drift and Pacifica are non-custodial and, for the most part, do not check identity, which places them outside the regulated perimeter entirely. For a US trader that cuts both ways: no one gates your access, but no one backs you up either, as Drift’s users learned in April. The North Korean angle sharpens the stakes, because a permissionless venue can double as a laundering surface, and the response runs through sanctions and anti-money-laundering enforcement, the machinery we examined in our look at Huione and Section 311. Regulated perps and permissionless ones are drifting toward two separate worlds, and a trader now has to pick which set of trade-offs to accept.

What comes next, and how to size up a venue

Several threads will decide how Solana perps look a year from now. Drift’s relaunch and its ability to actually repay claim-token holders will show whether an on-chain protocol can survive a nine-figure breach with its reputation intact. Pacifica’s eventual token launch will test how much of its volume was real demand and how much was airdrop farming. The real-world-asset perps that turned Hyperliquid into a venue for tokenized stocks and commodities are likely to spread to Solana, bringing fresh users and fresh regulatory friction. And the gap between regulated venues and permissionless ones will keep widening, pushing traders to decide which model they trust.

Whatever the venue, the same due diligence applies before you put money at risk:

• Look at open interest and protocol revenue, not just the volume headline, to judge whether an exchange is genuinely busy.
• Know who your counterparty is: a liquidity pool (Jupiter) or market makers on an order book (Drift, Pacifica).
• Check the oracle setup, since an imported price is a single point of failure at the moment of liquidation.
• Read how admin keys are managed, through multisigs, timelocks and audits, because the Drift hack was a key-management failure rather than a code bug.
• Understand the funding cost of holding a position, which rose alongside interest rates in 2026.
• Accept that you have no recourse if a venue breaks, and that any profit is still taxable.

Frequently Asked Questions

What is the biggest perp DEX on Solana?

Jupiter Perps is the largest by trading volume, handling roughly 80% of Solana’s perpetual-futures activity in mid-2026, with Drift Protocol and the newer, token-free Pacifica as its main challengers. Jupiter uses a peer-to-pool model, while Drift and Pacifica run order books.

Is Drift Protocol safe to use after the 2026 hack?

Drift lost around $285 million in an April 2026 admin-key breach, but it laid out a recovery plan backed by up to $127.5 million from Tether and other partners, kept its insurance fund intact, and issued claim tokens to repay affected users over time. It remains functional, though users should weigh that history and note that repayment is spread across future revenue.

How is a Solana perp DEX different from Hyperliquid?

Hyperliquid runs its own purpose-built blockchain with an order book inside consensus, while Solana venues sit on a shared, general-purpose network. That gives Solana exchanges deep composability and native oracles but exposes them to the chain’s congestion and shared risks, as the Drift hack showed when the stolen funds included another protocol’s pool token.

Are Solana perp DEXs legal for US traders?

Perpetual futures are regulated in the US by the CFTC, not the SEC, and permissionless Solana venues like Jupiter, Drift and Pacifica are non-custodial and generally do not check identity, placing them outside the regulated perimeter. US traders using them have no recourse if a venue fails, and any profits remain taxable.

What is JLP and how does it earn yield?

JLP is Jupiter’s liquidity pool, a basket of SOL, ETH, wrapped BTC and stablecoins that acts as the counterparty to every trade on the exchange. Holders earn most of the platform’s trading and borrowing fees plus the net losses of traders, which is real cash flow rather than token emissions, but they also carry the pool’s directional price risk.

Liam Brennan covers markets and on-chain derivatives for HOGE Wire.