How Morpheus Builder Subnets Work

Marketplace figures as of 3 June 2026; emission figures as of 4 June 2026. The ranking table further down refreshes daily from on-chain data.

Two kinds of subnet (scope of this piece)

Morpheus uses “subnet” for two distinct constructs. This piece is about one of them.

Builder Subnets are capital-formation pools on the BuildersV4 contract on Base. Stakers deposit MOR, the operator claims accumulated emissions from the 24% Community Builders pillar, and there is no on-chain enforcement of how those emissions reach stakers. This is what the article covers.

Compute Subnets are inference-provider listings on a separate contract: the Lumerin Diamond, deployed on Base (0x6aBE1d28…3030a) and Arbitrum. A Compute Provider runs a Morpheus-Lumerin Router node, stakes a small amount of MOR to lift their router ranking (current minimum: 0.2 MOR on Base, 0.1 MOR on Arbitrum), and earns MOR from the 24% Compute pillar based on session activity (Bid Price × Session Length, under the Yellowstone model). Third parties can stake into a Compute Subnet via the ProvidersDelegate contract and share the rewards minus an operator fee. As of 4 June 2026 there are 13 active Compute Subnet providers on Base and 19 on Arbitrum.

The two constructs share the word “subnet” and nothing else. Different contracts, different emission pillars, different mechanisms. When the Morpheus team talks about Compute Subnets on X or in the docs, they mean the second kind. When this piece talks about subnets without qualification, it means the first.

What the marketplace actually looks like

On the Mere Morpheus podcast the day after mordiem.com shipped, David Johnston called it “a real breakthrough to sort of simplify the way people understand that Morpheus can be this marketplace.” That’s the right frame for everything that follows.

Morpheus is a protocol that pays the entire builder pillar emission directly to whoever creates a subnet, then lets that operator decide what to offer back to stakers. As of 3 June 2026, 174 Morpheus builder subnets are live on Base, with around 1.91 million MOR staked across them.

Operators have done a variety of creative things with the same design surface:

• An official stake-for-access gateway that takes 74 stakers and sells them daily inference credits.
• A single-staker commercial inference node that the operator funds themselves.
• A yield bundle that pairs MOR emissionsEmissionsNew tokens created and distributed by a blockchain protocol over time as rewards to validators, stakers, or miners. Emissions fund network security and participation at the cost of diluting existing holders.Like a company that pays employees partly in newly printed shares. Every year the total number of shares goes up, which means existing shareholders own a slightly smaller slice of the same company unless the company grows faster than the printing.Read more → with reselling Venice DIEM inferenceInferenceRunning a trained AI model to produce an answer. Inference is what happens when you type a prompt into ChatGPT and get a response. The model takes your input, computes a best guess, and returns it.Like asking an expert for their opinion. The training was the decades they spent becoming an expert. The inference is the 30 seconds it takes them to answer your specific question.Read more → capacity over AntSeed.

The protocol is identical in all three cases. The product is whatever the operator chose to build on top of it.

This piece walks through what the Morpheus protocol gives operators to work with, what they are building with it today, and what a depositor should look at when picking which subnet to stake into.

I hold MOR (capital pool, six-year Power Factor lock) and stake MOR for inference access via the gateway. Disclosed.

What the builder rewards are for

Morpheus pays the Community pillar to get useful end-user products built on the protocol, and to bring users in behind them. The reward lands with the builder. The whitepaper is explicit: builders “earn MOR rewards by building applications for end users, including Smart Agents, front-ends, Dapps, and tools.”

Staking is how those emissions get directed. A MOR holder stakes toward the subnets they want funded, and the protocol routes that holder’s pro-rata share of the pillar to the builder they backed. In the whitepaper’s framing, holders “get to individually express which projects they think are most worthy of support.” The stake is a directed vote, and the builder is who it funds.

That reframes what a healthy marketplace looks like. Hundreds of subnets drawing emissions is the subsidy doing its job, not a yield being diluted. The question a staker is actually answering is which builders they want to fund and gain access to, rather than where the highest headline APY sits. Most subnets pay no MOR back to depositors, because direct staker yield was never the mechanism’s promise.

MOR is where the value is meant to accrue. Each project’s liquidity pairs against MOR, the way ETH sits under Ethereum projects, so a larger ecosystem of funded builders deepens demand for the base asset. The accrual target is MOR as the reserve asset of everything built on top, rather than any single subnet’s pass-through.

How emissions reach builder pools

MOR follows a strict linear emission curve. Day 1, on 8 February 2024, the network minted 14,400 MOR. Each subsequent day the daily emission drops by 2.468994701 MOR until it hits zero on day 5,833, around January 2040.

The cap is 42 million MOR. As of 4 June 2026 the network is at roughly day 848, emitting around 12,309 MOR/day. The curve is hard-coded in LinearDistributionIntervalDecrease.sol, with no governance lever to change it.

That daily emission gets split five ways:

• 24% Capital providers
• 24% Code contributors
• 24% Compute providers
• 24% Community builders
• 4% Protection Fund

Each pillar’s slice on day 1 is 3,456 MOR. With the linear decay applied, the Community (builders) pillar receives around 2,954 MOR per day as of 4 June 2026.

The builder pillar funds everything in this article. Annualised at the current rate the pillar is a touch over one million MOR per year, though a governance-set multiplier (covered in the next section) currently routes 80% of it to subnets. That’s the pool every operator is competing for a share of.

How much of the builder pillar reaches subnets

Only part of the builder pillar’s daily emission reaches subnets. Before any subnet is credited, BuildersV4 multiplies the bucket’s daily emission by a governance-set parameter called networkShare:

reward to subnets = raw bucket emission × networkShare

networkShare is currently set to 80% (on-chain value 8e24 against a precision constant of 1e25). At that setting, 80% of the 24% Community Builders pillar reaches subnets, so 19.2% of total MOR emissions reaches builder subnets, below the headline 24% allocation. The remaining 20% of the pillar, equal to 4.8% of total emissions, is withheld at the contract level and reaches no subnet.

The parameter is set by the contract owner, the Morpheus multisig (a 5-of-9 Gnosis Safe), which can move it anywhere between 0% and 100%. It does not track staking levels, and because it is capped at 100%, MOR withheld at a lower setting cannot later be released to subnets by raising the parameter.

networkShare is not described in any MRC, in the Morpheus gitbook, or in the node docs at nodedocs.mor.org. It is visible only by reading the deployed contract. Earlier builder contract versions (V1 and V2) carried no networkShare parameter at all.

What the protocol hands the operator

The single most important fact about the BuildersV4 contract is what happens when a subnet’s accumulated rewards get claimed.

The claim(bytes32 subnetId, address receiver) function, verified on the implementation contract at 0x18FAEf315b40a6d9cf49628f1133b1aa507513b0 on Base, transfers the full accumulated subnet rewards to whatever address the admin or claim admin specifies. The contract carries no internal logic for splitting rewards among individual stakers. That logic, by design, lives at the operator layer.

When an operator creates a subnet they pick:

• Name, slug, description, website, image
• Minimum deposit
• Withdraw lock period (measured from each deposit, not from a withdrawal request)
• The claim admin and receiver address that accumulated rewards are paid to

The V4 contract removed the Power Factor multiplier, virtual deposits, and the claim-lock that earlier builder versions carried. Rewards now track plain deposited MOR. (The Power Factor that still exists on the Capital pillar, MRC42, is a separate mechanism and does not apply to builder subnets.)

Stakers pick which subnet to deposit MOR into. Any pass-through of MOR to stakers, any service offered in exchange for the deposit, and any economic relationship beyond the bare on-chain deposited accounting is set up off-chain by the operator.

Fact: The protocol pays 100% of a subnet’s earned MOR to the operator’s chosen receiver, with no on-chain enforcement of distribution to stakers. The 7-day deposit lock is the only protocol-enforced staker protection.

Take: This creates a strong incentive mechanism for building on top of Morpheus. By stepping back from prescribing what an operator does with the emissions, the protocol opens the design surface for operators to:

• Pass MOR through directly to stakers
• Build a service on top of staked credit
• Run a partnership where stakers earn a partner token
• Gate access to a product

Stakers can read the off-chain offer, look at who’s running the subnet, look at the parameter choices on-chain, and decide whether the pitch matches what they want exposure to. The marketplace works because the protocol gets out of the way.

What MRC22 actually proposed

The marketplace as it runs today is broader than the design that seeded it. MRC22, the canonical builder-staking proposal, never assumed operators would pay MOR yield back to stakers. It assumed each builder would launch its own project token and reward stakers in that token instead.

The proposed mechanic: a MOR holder stakes to a subnet, the protocol directs that subnet’s MOR emissions into the project’s AMM pool, half is swapped for the project’s native token, and the result is added as protocol-owned liquidity. Stakers earn the project’s token, not MOR. In the words of the proposal, holders “get to individually express which projects they think are most worthy of support” and gain exposure to projects building on Morpheus without selling their MOR. MOR becomes the trading pair under every project’s liquidity, the way ETH sits under Ethereum projects.

So a builder subnet was designed as a launchpad. MOR-yield-straight-to-stakers, the pitch some subnets now advertise, was never the canonical model and is not something the contract does on its own. Where MOR does reach stakers, the operator is claiming the subnet’s rewards and passing a share back by hand, entirely off-protocol. That distinction is what a headline APY hides: the number is a promise from the operator, not a guarantee from the protocol.

The live ranking

172 subnets, ranked by MOR staked. This table reads from on-chain data refreshed daily. The top 10 sit below; the full directory, with claim-wallet type, withdraw lock and our own read on each subnet, lives on the Morpheus builder subnet directory.

Top 10 Morpheus Builder Subnets by MOR Staked

#	Subnet	Operator	TVL (MOR)	Stakers	Lock	Min
1	Mor.org-BASE	0x58a1…90c6	276,384	10	7d	0.001
2	MORDIEM Case study	0x515c…e1bc	198,933	7	7d	0.001
3	MOR-StakeVault-Base	0x2331…d104	170,265	21	7d	1
4	Lumerin Inference Node Case study	0x26b3…3f5e	160,170	2	9d	1
5	MySuperAgent	0x6776…96c1	143,393	26	30d	0.001
6	MOR Deployers	0x81d7…b768	111,132	2	7d	0.001
7	Morlord Shared	0x72f4…3be8	109,630	3	7d	<0.001
8	Morpheus Marketplace API Case study	0x504b…e28d	109,155	67	7d	0.001
9	Protection and Capital Incentive	0x6548…08ae	80,837	3	7d	0.001
10	MOR Yield Multiplier_Base	0x7768…6ea7	40,187	10	7d	1

A few things jump out of the current distribution:

• TVL concentration: as of 3 June 2026 the top 10 hold the majority of the 1.91 million MOR staked across the marketplace.
• Stakers vs TVL diverge: some subnets have small numbers of large depositors (the Lumerin Inference Node, with two depositors and around 160K MOR led by the operator’s own deposit). Others have broad participation (the official Morpheus Marketplace API gateway, with 70 stakers averaging around 1,600 MOR each).
• Parameter spread: minimum deposit varies from 0.000001 MOR to 1 MOR. Lock period defaults to 7 days but operators set it freely.

Three operator builds

The three highlighted subnets demonstrate three different approaches to what you can build on top of the Morpheus infrastructure. Each one is shipping today. None requires anything the protocol doesn’t already give every operator.

Stake-for-access (Morpheus Marketplace API)

The official subnet at app.mor.org is the simplest demonstration of the design.

Stakers deposit MOR into the BuildersV4 subnet. The contract converts each staker’s share of the total subnet stake into a daily inference credit allowance, refreshed at midnight UTC and scaled by the current MOR price. API calls debit credits. The MOR principal stays in the contract and remains withdrawable after the 7-day lock.

When a staker calls the API, the request routes via the Morpheus-Lumerin Router to a Compute Subnet provider on the Lumerin Diamond contract (separate from BuildersV4; see the disambiguation above). That provider earns MOR from the 24% Compute pillar via session activity, not from the Builder pillar that funded the staker’s credit allowance.

The Morpheus core team wears both hats here. They operate the Marketplace API on BuildersV4, capturing Community pillar emissions to fund the off-chain plumbing (rate-limiting, credit accounting, provider matchmaking). They also run two of the live Compute Subnets on the Lumerin Diamond as providers, both under the provider.mor.org:3333 endpoint. Two pillars flowing to one operator entity, each via its own contract.

As of 3 June 2026 the subnet holds around 112K MOR across 70 stakers, the broadest participation in the top 10. Stakers get the access they want without having to manage per-session payments to providers.

Operator-run capital pool (Lumerin Inference Node)

As of 3 June 2026 the Lumerin Inference Node sits at rank 3 with around 160K MOR staked across two depositors. The operator is the primary depositor, self-funding the pool.

Lumerin built the Morpheus-Lumerin proxy-router, the code the Morpheus network uses to route inference requests between users and providers. The proxy-router code is built on top of the existing Lumerin codebase from hashrate decentralisation, adapted for inference routing.

The framing trap here is to assume the roughly 160K MOR Builder Subnet stake is what powers Lumerin’s compute business. It is not. Lumerin separately operates a Compute Subnet on the Lumerin Diamond contract under a different address, with endpoint morpheus.lmn.lumerin.io:3333 and a small token stake. The Compute Subnet is what earns from inference work, paid from the 24% Compute pillar via session activity. The BuildersV4 pool we see in the top 10 does not route inference and does not earn from sessions.

So Lumerin’s build is one operator running both kinds of subnet under the same brand. The BuildersV4 stake is a Builder pillar capital pool, self-funded by the operator to capture a pro-rata share of Community pillar emissions. The Compute Subnet runs the actual inference node and earns separately. Two distinct contracts, two distinct emission streams, one entity wearing both hats.

Multi-pillar yield bundle (MORDIEM)

MORDIEM, deployed on 7 May 2026, is the newest of the three and the most cross-protocol. It is also more ambitious than it first appears. Ryan Condron is operating three interlocking products under the brand.

The three positions:

1. Builder Subnet on BuildersV4. As of 4 June 2026, around 193K MOR is staked across 7 on-chain depositors, though a single operator-controlled wallet holds roughly 99% of it. Offers a headline-level APY pass-through in MOR to stakers, funded from the subnet’s pro-rata Community pillar emissions. Stakers reach that offer through mordiem.com, which takes the deposit as a transfer to the operator’s wallet; staking into the BuildersV4 pool directly earns the staker nothing.
2. Compute Subnet on the Lumerin Diamond. As of 4 June 2026, around 2,285 MOR staked at endpoint mordiem.com (operator address 0xd01c1B…), the second-largest compute provider on Base. Earns MOR from the 24% Compute pillar via session activity. This is what the X post on Compute Subnets describes.
3. Cross-protocol Venice DIEM resale on AntSeed. DIEM holders stake their unused daily inference credit, MORDIEM contributes a Venice API key, AntSeed buyers pay USDC per inference request, and proceeds stream back to DIEM stakers (with an operator share set at the pool level).

Three Morpheus pillars are touched (Community Builders, Compute, and indirect MOR demand via the AntSeed flow). Three independent protocols are composed (Morpheus, Venice, AntSeed). One operator brand, no team-led integration required by any of the three protocols.

The two MOR products (Builder Subnet for MOR stakers, Compute Subnet for inference earnings) share a brand and operator but no on-chain economic flow between them. A staker backing the MOR position through mordiem.com is buying exposure to Community pillar yield; the Compute Subnet earns independently from session activity on the Lumerin Diamond. A staker depositing DIEM into the AntSeed pool is buying exposure to inference demand routed through a Venice API key.

MORDIEM is the demonstration the marketplace thesis needed. Four products composed by a fourth party into a single user-facing offering. Morpheus benefits twice (two pillars put to work), Venice benefits from DIEM credit demand, AntSeed benefits from a marketplace participant routing real volume. The design space allowed it, so an operator did it.

How to pick a subnet to stake into

A staker reading the table above is doing the same analysis any owner-side investor does on any cap-table: who’s running it, what’s the offer, what are the parameters, and what’s the realistic delivery risk. The protocol surfaces four on-chain signals and one off-chain offer to evaluate.

Fact: Subnet selection is a competitive marketplace question. As of 3 June 2026, 174 subnets are competing for around 1.91M MOR of deposits. The protocol surfaces the operator’s address, the subnet’s MOR stake, the staker count, the minimum deposit, the withdraw lock period, and the contract-stored description, slug, website, and image. Everything else is the off-chain offer.

Take: The most signal-dense thing to read is the gap between what the operator’s offer claims and what the on-chain parameters support. A 20% pass-through claim is mathematically supportable today because the builder pool is generously sized relative to current stake. That same claim becomes harder to sustain as the pool gets crowded with new subnets and every operator’s pro-rata share thins out.

Better questions for a depositor:

• What’s the operator’s plan when the pool’s competitive dynamics shift?
• Is there a service or relationship under the headline that survives the shift?

The off-chain offer is what differentiates one subnet from another at the same MOR-yield ceiling. Each of the three case studies above targets a different staker profile:

• Marketplace API offers credit allowance, a verifiable service with measurable consumption.
• Lumerin Inference Node offers exposure to the Builder pillar via a self-funded capital pool; the operator’s separate compute work runs on the Lumerin Diamond contract.
• MORDIEM offers MOR yield from the Community pillar plus indirect exposure to the operator’s Compute Subnet earnings and their cross-protocol Venice DIEM resale on AntSeed.

Withdraw lock, minimum deposit, and the receiver address are parameter signals worth reading:

• Long lock periods suggest the operator’s offer assumes commitment from stakers, a multi-week service relationship rather than a yield park.
• Fractional MOR minimums (down to one-millionth of a MOR) suggest the operator is targeting retail breadth, more stakers at smaller cheque sizes.
• The receiver address tells you where claimed MOR lands. A multisig or a contract receiver is a weak signal that distribution is structured; a single externally-owned wallet is the harvest-to-operator default.

Where this heads

Morpheus is in Zion, the current era focused on scaling for end users. The mechanism is live, the marketplace is competitive, the capital layer is multichain and multi-asset, and the operator design surface is wide open. The BuildersV4 contract pays 100% to the operator’s chosen receiver and leaves pass-through to the product layer.

Future contract-layer additions worth thinking about:

• On-chain commitments from operators to specific pass-through schedules.
• Standardised multisig or timelock defaults that operators can opt into.
• Dispute resolution for cases where an operator’s off-chain promise drifts from the on-chain flow.
• A formalised reputation surface that aggregates the off-chain offer history of an operator across multiple subnets they’ve run.

The protocol’s job today is to demonstrate that a permissionless marketplace can attract operators with diverse, differentiated products. 172 subnets and the three builds visible in the top 10 are the demonstration. Marketplaces work when the protocol gives operators what they need (predictable emissions, on-chain accounting, settled deposits) and trusts them to build the product on top.

Morpheus pays builders generously in MOR. What they do with that opportunity, and how they earn the trust of stakers who fund their work, is the part the protocol leaves open by design. That’s the marketplace breakthrough Johnston was pointing at.

---

Related reading: How MOR Actually Works walks through the capital-provider side of the same protocol. The Morpheus project review covers the full Freedom and Returns scoring. The DIEM-Venice perpetual AI credit piece explains the mechanic MORDIEM is reselling, and the AntSeed review covers the marketplace MORDIEM routes through.