Solana Confidential Compute launches private AI toolkit on devnet

Solana Confidential Compute launches private AI toolkit on devnet

The Solana Foundation and private AI startup Nito launched Solana Confidential Compute on Nov. 13, 2026, Crypto Briefing reported. The open-source toolkit is designed to help developers build private AI applications and verify their computing hardware on Solana, according to Crypto News.

According to Crypto News, the AI work is performed away from the blockchain, but applications can examine records on Solana that verify the hardware used. Those records can be published without also disclosing a user's conversation or private inputs.

The tools are an unaudited beta on Solana devnet and are intended for developers building and testing applications, according to Crypto News. A mainnet version is planned, but the announcement did not give a release date, the outlet reported.

How Solana Confidential Compute verifies hardware

The release has two main components: a software development kit and an on-chain verification program, Crypto Briefing reported. Kucoin reported that Nito is integrating the infrastructure into its private AI platform, while the open-source components are also available for reuse by other teams.

Crypto News reported that the system uses trusted execution environments, or TEEs, to keep code and data separate from the rest of the computing infrastructure, which includes the host operating system and the person operating it. Crypto Briefing described an attestation as a hardware-signed statement about what is running inside such an environment.

Inside a confidential virtual machine, software generates a signing key that stays within the protected hardware. A report signed by the hardware connects the key with a recent Solana slot. The verification program examines that report, then stores the signing key alongside the hardware type and expiration date.

According to Crypto News, the machine uses that key to sign the answer it produces. The outlet reported that an application receiving the answer can check the signature and consult the corresponding Solana record before accepting the result.

The beta supports attestations from Intel TDX and AMD SEV-SNP, while NVIDIA GPU attestation remains in development. The on-chain program also lets people see whether hardware passed a verification check, Crypto Briefing reported.

The private chat example

A developer chat template shows how Solana Confidential Compute can be used in an application built with Next.js and React, according to Crypto News. The outlet reported that the template's documentation calls for the browser to verify the confidential environment before encrypting a message to its key.

If verification fails, the prompt is not sent, Crypto News reported. In the documented flow, encrypted messages travel first through the application server and then Nito before reaching the protected environment, where a model reads the prompt and produces a response.

Replies are encrypted inside the TEE before they return to the browser. Under the documented design, neither the application server nor Nito can read the message contents, even though messages pass through both.

The encryption does not conceal every detail of a request. According to Crypto News, the template documentation says a visitor's IP address is visible to the application server; Nito can observe which account and model are used, when a request is made and how large the encrypted message is.

Developer and security limits

Crypto News reported that developers using Solana Confidential Compute continue to depend on the code in the webpages they serve, the manufacturer's hardware attestation and the published software operating within the protected environment. Crypto Briefing likewise reported that TEE privacy depends on trust in the chipmakers whose hardware produces the attestations.

Verification can identify a published software release, but it does not establish that the release contains no bugs, according to Crypto News. Kucoin reported that the process also does not guarantee the safety of the webpage code a developer serves.

Developers must implement authentication, rate limiting and bot protection themselves, and the example flow requires a funded Nito account and an API key. The Foundation has described the release as unaudited and intended for development, Crypto News reported.

Applications and mainnet plans

The Foundation has identified personal assistants, agents handling confidential business data and private trading strategies as potential applications, according to Crypto News. The outlet also reported that the announcement listed AI rulings for prediction markets or insurance claims and AI governance decisions as possibilities.

Another possibility in the announcement is trading orders accepted only from verified machines, Crypto News reported. The Foundation told developers in its X thread that they could verify a signing key's attestation status before accepting results bearing its signature, Crypto News reported.

Nito is incorporating the infrastructure into its private AI platform, while making the same open-source components available for other teams to use, according to kucoin. Crypto Briefing reported that Nito runs open-weight AI models inside attested hardware to allow end-to-end encrypted interactions.

For Solana Confidential Compute, the next planned network release is mainnet. Crypto Briefing reported that it is expected after a security audit, with no timeline provided for either the audit or the launch.

Conclusion

The Solana Foundation and Nito have released an open-source toolkit that combines private AI processing with hardware verification on Solana, according to Crypto News. The processing happens off-chain, and a verification record can be published without exposing a user's private inputs. The release is unaudited and intended for development, the outlet reported. A mainnet version is planned, with Crypto Briefing reporting that it is expected after a security audit. No timeline has been given for the audit or the launch, according to Crypto Briefing.

Frequently Asked Questions

When did Solana Confidential Compute launch, and where is it available?

The Solana Foundation and Nito launched Solana Confidential Compute on Nov. 13, 2026, Crypto Briefing reported. According to Crypto News, the toolkit is an unaudited beta on Solana devnet for developers building and testing applications. A mainnet release is planned, but no date has been announced.

Which hardware does Solana Confidential Compute support?

Intel TDX and AMD SEV-SNP are covered by the beta; work on NVIDIA GPU verification has not finished. The system checks a hardware-signed report that ties a key to a recent Solana slot, then records the key, hardware type and expiration date on-chain.

What can Nito and application servers see in Solana's AI chat template?

The documented design keeps message contents unreadable to both intermediaries. Crypto News reported that Nito can observe which model and account are used, the time of a request and the size of its encrypted message. The application server, meanwhile, can view the visitor's IP address.

When will Solana Confidential Compute reach mainnet?

Crypto News reported that a mainnet version is planned but has not been assigned a release date. The toolkit is currently on devnet in a beta intended for developers building and testing applications. It has not yet been audited, according to the outlet. Crypto Briefing said an audit is expected to precede the launch, but neither has a timetable.

Sources

Read Also

AKT migration to Solana planned through one-way portal

AKT migration to Solana planned through one-way portal

Akash has selected Solana as the destination for its network and outlined a one-way AKT migration from its existing chain. Governance Proposal 346 ratified the decision, with 97% of participating token holders voting in favor, Cision PR Newswire reported.

The move separates the token transfer from the later relocation of Akash's compute marketplace. According to Akash, tokens sent to Solana cannot be moved back through the migration system, while live workloads remain unchanged until a marketplace cutover that has not yet been dated.

How the AKT migration will work

Akash says it plans to launch AKT as an SPL token on Solana and open an AKT migration portal for two years, two weeks after the proposal passes and only after two independent audits have passed. Cision PR Newswire reported that holders will be able to transfer tokens at a one-to-one ratio.

The token will retain its AKT ticker and six decimal places, according to Akash. The team says tokens sent to the portal leave circulation on Cosmos, so the transfer does not change total supply.

Akash's AKT migration instructions describe a portal flow in which a holder connects Keplr, enters a Solana address and an amount, and confirms the transfer. The team also describes a command-line alternative and says the public portal URL will be announced separately.

According to Akash, a module on the original chain locks the tokens and records each transfer, after which a relayer submits block headers and proof to Solana. Akash says its Solana program checks that proof, and independent reporters then check the result and vote on the Akash chain.

If Solana cannot deliver a transfer to the specified address, it records a void and the AKT follows a refund process on Akash, the team says. Akash says a program address or token account, rather than a wallet address, can cause a void, and the network fee for the deposit is not returned.

Akash governance can set and change transfer limits, according to the team. Akash says minting on Solana can begin when deposit proof is available, but affected mints must wait until the next UTC day if Solana's separate daily cap has been exhausted.

Cutover, staking and live leases

The marketplace will move later, with new deployments shifting to Solana once audits and rehearsals pass, according to Akash. A separate governance vote must set the cutover date, and the plan also calls for a sunset upgrade; Akash says workloads will keep running for 90 days after cutover while tenants redeploy.

Holders who want to undertake the AKT migration with staked tokens must first unbond them, a process Akash says takes 21 days. Until cutover, the team says staked tokens continue to secure the old chain, earn rewards and vote on remaining proposals; staking rewards on that chain end at cutover.

Founder Greg Osuri says staking on Solana will not be live for several months, according to TokenPost and Yahoo Finance. Akash says it has yet to settle how unvested AKT will migrate, how tokens still on Cosmos at the chain halt will reach the portal, and what happens to tokens not moved within two years.

Before cutover, providers must upgrade a daemon, re-register on Solana with collateral and be re-audited, according to Akash. The team says old-chain earnings and refunds of deposits made before cutover will be paid weekly in new AKT, while top-ups made after cutover will not be credited on Solana.

Akash says its updated chain SDK will be a breaking release built on Solana tooling while retaining the high-level API. The team says its Console API will use Solana addresses in version 2, with version 1 frozen on old-chain data, and that wallets, decentralized exchanges and custodians can add SPL AKT before the marketplace moves.

Why Akash selected Solana

Akash says it sought a shared-security provider among established layer-one networks and assessed proposals against goals that included reducing the AKT staking burden, cutting operational overhead and preserving interoperability. Of seven options considered, Solana mainnet and an existing Ethereum layer-two network met every hard requirement, according to Akash.

Overclock Labs published a 15-document plan detailing both a Solana route and an Ethereum layer-two route, Akash says. The team says it favored Solana on factors including cost under load, the suitability of its runtime for the marketplace and the presence of comparable projects.

Akash says about 100 validators currently bond AKT to secure consensus, while the core team maintains forks of cosmos-sdk, CometBFT and gogoproto. The team argues that the network bears the cost of its own chain without using its sovereign powers, and says Solana programs inherit security without requiring Akash to bond tokens or stake SOL.

At current volume, Akash estimates network fees on Solana at about 0.075 SOL a day. The team puts its marketplace activity at roughly 15,000 transactions daily, compared with 112 million daily user transactions on Solana, and says its throughput can scale with the block because each lease has its own account.

Akash cites Helium and Render as examples of projects that have migrated to Solana. It says Render's token upgrade portal is a model for its own transfer process and that Helium's migration tooling is available for reuse.

Risks and remaining decisions

Akash acknowledges that the move gives up sovereign consensus, validator income and native IBC. The team also says it cannot control Solana's block times or fee markets, and that economic slashing is not yet live on Solana.

Solana experienced several full outages between 2021 and February 2024, according to Akash. The team says Solana nevertheless met its requirement for 99.5% liveness over the preceding 24 months.

Akash identifies the migration itself as the source of the largest risks and says each has a mitigation and a gate to clear. The team says the Solana Foundation responded to its request by offering funding, assistance with audits, a design for subsidizing gas and a phased rollout plan.

Cision PR Newswire reported that Akash Marketplace, AkashML and Akash Homenode are to be available natively on Solana. For the marketplace, Akash says the next operational step remains dependent on audits and rehearsals, followed by a governance decision on cutover.

Conclusion

Akash has chosen Solana and received community approval for the move, according to Akash and Cision PR Newswire. Its plan separates a one-way token transfer from the later marketplace cutover: the portal is scheduled to open only after the specified waiting period and two independent audits have passed. Akash says the marketplace date requires a separate governance vote after audits and migration rehearsals, while the treatment of tokens left on Cosmos at the chain halt remains unresolved. The old chain is scheduled to halt 90 days after cutover, according to the team.

Frequently Asked Questions

When will the AKT migration portal to Solana open?

Akash says the portal is planned to open two weeks after approval of the migration proposal, provided two independent audits have been completed successfully. It is intended to remain open for two years. Cision PR Newswire reported that the proposal received community approval; Akash says the public portal URL will be announced separately.

Can AKT holders move tokens back from Solana to Cosmos?

No. Akash says its migration system transfers AKT in one direction and cannot send tokens that arrive on Solana back to the Akash chain. Cision PR Newswire reported that the portal will support transfers at a one-to-one ratio. Akash says the token will keep its ticker and six decimal places.

How do holders migrate staked AKT to Solana?

Akash says holders must unbond staked AKT before using the migration portal, and that unbonding takes 21 days. Until the later marketplace cutover, staked AKT continues to secure the old chain, earn rewards and vote on remaining proposals, according to the team. Akash says old-chain staking rewards end at cutover.

When will the Akash chain halt after the Solana cutover?

Akash says the old chain will halt 90 days after the marketplace cutover. The team has not set a cutover date: it says a separate governance vote will establish one after audits and rehearsals pass. Akash says live workloads continue during the 90-day period while tenants redeploy.

Sources

Read Also

Solana validator count could fall under changes, kucoin reports

Solana validator count could fall under changes, kucoin reports

The Solana Foundation has changed its validator delegation program, introducing a term of 12 months and limiting participation to 100 validators, kucoin reported. Jacob Creech, the foundation's vice president of technology, said reducing the Solana validator count was not the aim, according to the outlet. The foundation instead wants to direct support toward network health and key ecosystem contributors, kucoin reported.

The changes have prompted community concerns that validator numbers could fall, kucoin reported. Creech said a decline could nevertheless occur if there were not enough high-quality contributors, while more than 100 validators have applied for continued support, according to the outlet. His response addressed the potential effect on validator numbers, rather than presenting a reduction as an objective of the program changes.

Solana validator count and program limits

Under the revised Solana Foundation Delegation Program, participants must operate their own validator nodes, kucoin reported. They also automatically graduate once 300,000 SOL is staked, according to the outlet. The staking figure is the threshold for automatic graduation, while the separate node requirement concerns how participants operate, according to kucoin.

The 100-validator cap applies to the delegation program, according to kucoin. The concern Creech addressed was whether the changes could affect the Solana validator count more broadly, the outlet reported. The cap sets a limit on program participation, rather than a stated ceiling on the validator count across the network.

More than 100 validators have submitted applications seeking continued support, kucoin reported. That figure is higher than the program's stated cap of 100 validators, according to the figures reported by the outlet. The two figures describe different things: one is the number seeking continued support, while the other is the program's participation limit. The applications do not, by themselves, establish how many validators will receive support.

The program's numerical thresholds serve different purposes, according to kucoin: the 100-validator figure caps participation, while 300,000 SOL is the staking threshold for automatic graduation. The 12-month term is a separate condition, as is the requirement for a participant to operate its own node, the outlet reported.

Original program objectives

Creech said the delegation program was originally built around three objectives, according to kucoin. These were helping new validators launch, establishing a healthy testnet and enabling selected validators to pre-test network updates, the outlet reported. Two of those objectives concerned the testnet and update testing, while the other focused on helping new validators begin operating.

As the Solana ecosystem expanded and its validator community matured, the original mechanism became unable to fully meet the network's changing needs, kucoin reported. The outlet said the foundation now wants a more proactive and flexible delegation model. The proposed approach places its focus on network health and key ecosystem contributors, according to the report.

That proposed model is intended to support network health and key ecosystem contributors, according to kucoin. In addressing concerns about the Solana validator count, Creech distinguished that aim from a possible incidental decline in validator numbers, the outlet reported. His explanation left open the possibility of a decline if enough high-quality contributors cannot be found, even though lowering the count is not the goal.

Future delegation support

The foundation plans to keep supporting new validators through the program, kucoin reported. It also intends to use delegation outside the program to reward validators making substantial contributions to network health, according to the outlet. That separates the continuing support for new validators from another route for rewarding contributions through delegation.

The foundation has previously delegated stake to ecosystem contributors including Blueshift, kucoin reported. It has also supported validators that were among the first to test Frankendancer, according to the outlet. Those earlier examples involved support for ecosystem contributions and for validators participating in testing, two areas related to the foundation's stated objectives.

The original program included help for new validators and pre-testing of network updates among its objectives, Creech said, according to kucoin. Under the intended approach, new-validator support remains part of the program, while delegation outside it can reward contributions to network health, the outlet reported. The foundation has also said it wants the broader delegation model to be more flexible, according to kucoin.

That planned support for contributors sits alongside the program's continued assistance for new validators, kucoin reported. Creech nevertheless said the Solana validator count could decline if enough high-quality contributors could not be found, though he did not describe a reduction as the objective, according to the outlet.

Conclusion

According to kucoin, the Solana Foundation has set a 12-month term and a 100-validator limit for its delegation program, alongside a node requirement and a 300,000 SOL graduation threshold. Creech said a lower validator count was not the objective, although it could occur if there were too few high-quality contributors, the outlet reported.

The next stated step is to continue supporting new validators through the program while using separate delegations to reward those making substantial contributions to network health, according to kucoin. The foundation also intends to make its delegation model more flexible, the publication reported.

Frequently Asked Questions

Why could the Solana validator count decline?

Creech said validator numbers could fall if there were not enough high-quality contributors, according to kucoin. He said that would be an incidental outcome rather than the goal of the changes, which prompted community concerns about their potential effect on validator numbers, the outlet reported.

What are the Solana delegation program's 12-month and 100-validator limits?

The foundation has set a 12-month participation term and capped its delegation program at 100 validators, kucoin reported. The outlet also reported that more than 100 validators had submitted applications seeking continued support, a figure higher than the program's stated cap.

How does 300,000 SOL affect Solana validator participation?

Participants automatically graduate from the delegation program once 300,000 SOL is staked, according to kucoin. The outlet also reported that validators must operate their own nodes under the revised guidelines, alongside a 12-month term and a cap on program participation.

What support will the Solana Foundation give validators outside the program?

The foundation plans to reward validators that make substantial contributions to network health through delegation outside the program, kucoin reported. It will also continue supporting new validators within the program, according to the outlet, as part of a more flexible approach to delegation.

Sources

Read Also

Aave stablecoin yields can fall after Fed tightening, paper finds

Aave stablecoin yields can fall after Fed tightening, paper finds

A Banque de France working paper published on November 9 found that Aave stablecoin yields can fall immediately after an unexpected Federal Reserve tightening, Coindoo reported. The study tracks deposits of USDC, USDT and DAI on the lending protocol from January 2021 to January 2026.

The finding matters because, as Genfinity reported, a growing share of dollar savings earns rates that the Fed influences only indirectly. Genfinity also reported that centralized exchanges increasingly use protocols including Aave and Morpho to offer yield on customer balances.

The paper is by Andrea Barbon, Jean Barthélemy and Benoît Nguyen, according to Coindoo. The outlet reported that the authors link the initial rate decline to leveraged borrowers repaying faster than suppliers withdraw, reducing the share of deposits being borrowed.

Aave rates during Fed moves

The rate history examined in the paper includes periods when Aave stablecoin yields moved in the opposite direction from Fed policy. Between March 2022 and July 2023, the Fed raised rates by a combined 525 basis points, while the average Aave deposit rate fell and then returned to its starting point.

The average Aave rate dropped from 2.5% to 0.5% by September 2022. It had recovered to 2.5% by July 2023, even as the Fed's increases over the period totaled 525 basis points.

A different divergence appeared in late 2024. From September through December, the Fed cut rates by 100 basis points, while Aave rates rose from about 3% to 9%.

The longer sample also contained periods when DeFi rates were above and below the Fed funds rate. Coindoo reported that the gap averaged roughly 100 basis points above the Fed rate across the sample, but DeFi rates were about 160 basis points below it from mid-2022 through the end of 2023.

How Aave stablecoin yields are set

According to Coindoo, suppliers deposit stablecoins into Aave pools, and borrowers draw from those pools against crypto collateral. The proportion of deposited tokens borrowed is the utilization rate, a measure that helps determine what lenders earn.

Coindoo reported that Aave's rate model raises the cost of borrowing as utilization increases, with the rate rising more steeply beyond a threshold. After a reserve factor is retained for the protocol, the remaining interest goes to suppliers, according to the outlet.

For USDC, the paper lists parameters as of March 2025 under which the borrowing rate rose from zero to 4% at 92% utilization, Coindoo reported. Above that threshold, the listed slope was steeper by 10 percentage points, and the reserve factor was 10%.

The model makes the relative movement of borrowing and supply important. Coindoo reported that when borrowing shrinks while supply stays in place, utilization and the supplier rate fall; if both sides shrink at the same pace, utilization and the rate do not change.

Testing policy surprises

The paper's main test focuses on monetary-policy surprises rather than every Fed rate decision, according to Coindoo. The outlet reported that the authors measured surprises using interest-rate futures in the half hour around Federal Open Market Committee announcements. The authors then standardized each surprise to correspond to an increase of 25 basis points in the yield on two-year Treasury securities, the outlet reported.

Both total borrowing and total supply reached declines of around 10% four to five months after a shock of that size. Borrowing fell more quickly at the outset, and Aave stablecoin yields dropped immediately across all three assets, Coindoo reported.

Coindoo reported that the estimated effect subsequently moved toward a slightly positive level that the authors could not statistically distinguish from zero. Because the test concerns surprises, the outlet said the result should not be treated as establishing the same effect for a widely expected hike.

Coindoo reported that Aave borrowers are predominantly leveraged long on crypto and that Bitcoin fell substantially and persistently after a tightening shock in the paper's sample. Genfinity likewise reported that traders tend to reduce leverage when the Fed tightens and Bitcoin falls.

Long-run rates and lending scale

The paper does not conclude that Aave stablecoin yields and traditional rates remain apart indefinitely. Genfinity reported that the authors' statistical tests found deposit rates move toward risk-free rates over the long run, although the leverage channel can delay that convergence after a policy surprise.

Coindoo reported that, in the authors' calculations, a shock affecting the difference between SOFR and Aave deposit rates took about five weeks to halve for DAI, four weeks for USDC and just over two weeks for USDT. Those figures describe how quickly a shock to the gap diminished, rather than the size of a deposit yield.

Genfinity reported that the authors also identified a shortage of large, active arbitrageurs connecting DeFi and traditional markets. According to the outlet, the paper said that gaps consequently last longer than they would in a conventional money market.

The study comes as Aave deposits have exceeded $50 billion since August 2025, according to Genfinity, which reported that the paper compares that scale with small U.S. deposit-taking banks. Genfinity also reported that Banque de France working papers do not represent the official views of the bank or the Eurosystem.

Conclusion

The working paper documents cases in which Aave deposit rates moved against Fed policy, including a decline during the 2022 tightening cycle and a rise during the 2024 cuts. Its test of unexpected tightening found that borrowing initially fell faster than supply, lowering rates on the three stablecoins studied, Coindoo reported. The subsequent stage of the authors' analysis follows the effect over the year: according to Coindoo, it moved toward a slightly positive level that could not be statistically distinguished from zero. The reported findings concern policy surprises and should not be assumed to apply in the same way to an expected hike.

Frequently Asked Questions

What did the Banque de France paper find about Aave stablecoin yields?

The paper found that Aave deposit rates can fall immediately after unexpected Fed tightening, according to Coindoo. The outlet reported that leveraged borrowers repay faster than suppliers withdraw, reducing utilization and the rate paid to stablecoin lenders. The study covers USDC, USDT and DAI from January 2021 to January 2026.

How did Aave deposit rates change during the Fed's 2022 tightening cycle?

The average Aave deposit rate fell from 2.5% to 0.5% by September 2022, then recovered to 2.5% by July 2023. Between March 2022 and July 2023, the Fed raised rates by a combined 525 basis points.

Why can an unexpected Fed hike lower Aave lending rates?

Coindoo reported that an unexpected tightening can prompt leveraged borrowers to repay faster than stablecoin suppliers withdraw. That lowers utilization, the share of deposits borrowed, and can reduce the rate suppliers earn. Aave's model raises borrowing costs as utilization rises and increases them more steeply beyond a threshold, according to the outlet.

Who wrote the November 9 Banque de France working paper?

Andrea Barbon, Jean Barthélemy and Benoît Nguyen wrote the paper, according to Coindoo, which reported that it was published on November 9. Coindoo also reported that the paper expresses the authors' views rather than those of the Banque de France. Genfinity said working papers do not reflect the official view of the bank or the Eurosystem.

Sources

Read Also

AAVE price article requires sourced facts

AAVE price article requires sourced facts

Sourced facts with IDs and corroboration counts, along with any exact quotes, are needed to report the AAVE price decline and comparisons with Bitcoin and Ether. The supplied facts and quotes lists are empty, so a longer article cannot be substantiated.

Sources

Read Also