Proposal #343
Passed on 21 Oct 2026 with 98.9% of the votes for yes.
Yes
98.9%
100.0% without abstain
Veto
0.0%
of votes cast
Ended
21 Oct 2026
Asked
299.7K
AKT · 7.1% of the pool
98.86%
of votes cast
1.14%
of votes cast
0.00%
of votes cast
0.00%
of votes cast
This proposal seeks community approval to fund the Overclock Labs core team for the protocol and provider-level engineering delivered during Q3 2026. Building on the Q2 work funded in [Hidden link], this work spans the Pyth price feed migration and resilience work behind BME, a unified Akash CLI, ARM64 provider support, the mainnet release of confidential compute and InfiniBand interconnect, and a single provider configuration file.
The AEPs and initiatives covered by this proposal are:
Akash's supply has been limited by assumptions built into the protocol. Every machine was assumed to be amd64, so providers with ARM hardware had no way to advertise it, and an amd64 container placed on an ARM node would fail after the tenant had already started paying for the lease. ARM64 support removes that limit and makes room for ARM servers and Apple Silicon. Confidential compute and InfiniBand interconnect support, whose core development was funded in Q2, were released to mainnet this quarter, giving tenants hardware-isolated workloads and multi-node GPU clusters for distributed training and fine-tuning.
The Pyth migration (AEP-88) is protective work. The on-chain oracle and Burn Mint Equilibrium both depend on an accurate AKT price, and the guardian-based Pyth feed Akash relies on is being retired upstream. The migration ran the old and new sources side by side before cutover so the price feed never went dark. The price feed resilience work is a direct continuation of that hardening, aimed at making the price feed layers behind BME sustainable rather than dependent on a single provider.
The CLI (AEP-89) and provider configuration work address the day-to-day experience of the people building on and supplying the network. Commands spread across separate binaries could look and behave differently depending on the tool, and provider settings split between CLI flags and environment variables made misconfiguration easy. Consolidating both reduces support load, simplifies documentation and automation, and makes onboarding new providers easier.
For the Q3 2026 Core Engineering Proposal, Overclock Labs is requesting $205,916 from the Community Pool for work directly attributable to the core AEPs for this period, alongside a 54.3% AKT volatility buffer.
| Request from CP | |
|---|---|
| Engineering hours | 726.71 |
| Engineering rate per hour | $176.90 |
| Engineering total cost | $128,553 |
| Support hours | 33.55 |
| Support rate per hour | $146.01 |
| Support total cost | $4,898 |
| Labor Subtotal | $133,452 |
| AKT Volatility Buffer (54.3%)* | $72,464 |
| Total Request | $205,916 |
*AKT Volatility Buffer – A 54.3% buffer is applied to this request to ensure adequate coverage during liquidation based on the past 180 days of AKT price fluctuations (applied as of 10/1/26). Any unused buffer during liquidation will be returned.
| Hours | Cost | |
|---|---|---|
| Pyth Router-Based Price Feed Contracts (AEP-88) | 209.12 | $41,417.74 |
| As a percentage of total labor budget | 27.51% | 31.04% |
| Unified Akash Command-Line Interface (AEP-89) | 162.71 | $28,298.33 |
| As a percentage of total labor budget | 21.40% | 21.20% |
| ARM64 Support | 80.34 | $16,178.81 |
| As a percentage of total labor budget | 10.57% | 12.12% |
| Confidential Compute (AEP-83) – Mainnet Release | 130.56 | $16,280.82 |
| As a percentage of total labor budget | 17.17% | 12.20% |
| InfiniBand / RoCE Interconnect – Mainnet Release | 38.58 | $7,475.25 |
| As a percentage of total labor budget | 5.07% | 5.60% |
| Price Feed Resilience | 86.11 | $17,649.74 |
| As a percentage of total labor budget | 11.33% | 13.23% |
| Provider Configuration (non-AEP) | 52.84 | $6,151.07 |
| As a percentage of total labor budget | 6.95% | 4.61% |
| Total | 760.26 | $133,451.76 |
Q3's core work widens what the network can offer and keeps its foundations sound. ARM64 support brings a new class of hardware onto Akash, and the mainnet releases of confidential compute and GPU interconnect make the network viable for sensitive and distributed training workloads. The Pyth migration kept the price feed behind the oracle and BME running through an upstream change, and the resilience work reduces its dependence on a single source. The CLI and provider configuration work make Akash easier to build on and easier to supply.
Overclock Labs will custody the requested funds in a new, distinct wallet so that funds from any other source are not commingled.
All funds will be liquidated and managed to minimize market impact. These funds will be handled with the same care and attention as all previous Community Funding Proposals, with liquidations done in a fashion that will not adversely affect the market. In practice, the effort of this liquidation will add depth to the AKT market for buyers looking to enter.
Overclock Labs will coordinate with the Decentralized Cloud Foundation (DCF) on the timing and execution of liquidations, so that conversions of these funds are aligned with other market activity across the ecosystem and are not working against one another.
Validators and delegator accounts, counted one vote each.
43 of 79 validators voted, 91% yes. 1 delegator accounts voted on their own, 100% yes.
Each validator's vote, by voting power.
43 validators voted, together 56.1% of the voting power.
