Switchboard’s September 25 support deadline has become an important test for Solana applications that historically relied on its oracle infrastructure.
The oracle provider announced on September 19 that its core contributor, Switchboard Technology Labs, would wind down and that existing implementations were being deprecated. Projects were urged to migrate to alternatives such as Pyth and RedStone.
However, one major question remains difficult to answer: which live Solana markets were still relying on Switchboard when support officially ended?
Public documentation shows that Switchboard has been integrated into several major Solana protocols, but documentation alone cannot confirm whether those integrations remained active on September 25.
Switchboard Technical Support Ends on September 25
Switchboard said existing technical support would end on September 25 after announcing its wind-down six days earlier.
The announcement represented a clear operational deadline, but it did not necessarily mean every Switchboard price feed stopped updating immediately.
Oracle systems can continue operating through independent infrastructure, existing gateway operators, or protocol-specific fallback mechanisms even after the original development team stops providing support.
This means the real impact depends on how each Solana application configured its oracle system.
Historical Integrations Do Not Prove Current Dependence
Switchboard has previously listed several well-known Solana projects as users, including:
- Kamino
- Jito
- marginfi
- Drift
However, those names represent historical integrations rather than a real-time list of active feeds.
A protocol may still support Switchboard in its code while no longer using it for live markets.
Likewise, a project may have migrated some markets while keeping Switchboard active in others.
This makes it important to distinguish between software support and actual production use.
Three Levels of Oracle Evidence Matter
There are three different types of evidence when trying to determine whether a protocol still relies on Switchboard.
The first is a marketing or integration page showing that the protocol previously used the oracle.
The second is technical documentation showing that the software is capable of reading Switchboard feeds.
The third, and most important, is the live configuration of an active market or vault.
Only the live configuration can confirm whether a particular market was actually depending on Switchboard at a specific time.
Even then, a backup oracle may be configured, which can reduce the impact of a primary feed going offline.
marginfi Has Already Built Alternatives
Marginfi offers one of the clearest examples of a protocol preparing for oracle migration.
Its documentation still lists SwitchboardPull among the oracle types its software can understand.
But that does not mean every marginfi bank still uses Switchboard.
A more recent marginfi upgrade introduced nine new oracle configurations that do not depend on Switchboard.
These include Kamino Scope feeds and alternative pricing methods for certain liquid staking and principal tokens.
The change shows that marginfi created a technical path away from Switchboard before the support deadline.
Older marginfi Software Could Face a Different Problem
Marginfi’s oracle migration also introduces another potential issue.
The protocol warned that older SDK versions may not understand the newly added oracle configuration types.
In some cases, a bank using one of the new oracle values could prevent an older client from initializing properly.
That means a migration away from an unsupported oracle can solve one infrastructure problem while creating another compatibility issue for applications that have not updated their software.
Marginfi advised developers to use SDK version 2.8.0 or newer.
Jito Documentation Still Mentions Switchboard
Jito’s Tip Router documentation presents a different picture.
Its public documentation still names Switchboard as a pricing source used to calculate the relative weights of assets such as JitoSOL and JTO.
The system uses these prices to help determine reward distribution across vaults.
However, the documentation also describes backup weights that can be used when oracle feeds become unavailable.
This means a Switchboard outage would not automatically imply that the entire Tip Router system stops working.
Jito’s Documentation May Be Outdated
There is an important limitation.
The Jito Tip Router documentation referenced in the article had been last updated around nine months earlier.
That means it can confirm how the system was designed, but it cannot prove the current live configuration.
Jito may have changed its oracle accounts without updating the documentation.
Alternatively, it may still use Switchboard while relying on fallback pricing when necessary.
Without examining current onchain accounts or receiving a recent statement from Jito, the active dependency cannot be verified.
Oracle Code Can Remain After a Migration
Jito’s public code also contains logic for retrying Switchboard oracle gateways.
But code alone does not prove that every active vault still depends on Switchboard.
Software often keeps compatibility code long after production systems have migrated to another provider.
The more important question is whether recent live transactions are still updating Switchboard accounts that are used by active vaults.
Not Every Oracle Failure Has the Same Impact
Different protocols use oracle prices for different purposes.
In lending markets, price feeds may determine:
- Collateral value
- Borrowing limits
- Liquidation thresholds
- User health calculations
In a reward system such as Jito’s Tip Router, oracle prices may instead affect how assets are weighted for distribution.
These are very different risk profiles.
A stale oracle in a lending market could potentially affect borrowing and collateral calculations, while a stale feed in a rewards system could instead affect allocation weights.
That is why counting protocol names alone does not accurately describe risk.
Kamino Scope Adds Another Layer
Kamino’s Scope system makes the dependency question even more complicated.
Scope is an onchain price aggregator rather than a single oracle provider.
It can collect values from multiple oracle sources and place them into one account.
A downstream protocol may therefore appear to use Scope, while the actual upstream data may come from Pyth, Switchboard, or another provider.
Seeing Scope in a market configuration does not by itself identify the true oracle source.
Scope Requires Source-Level Verification
Marginfi’s September upgrade lists new Scope-based configurations that do not depend on Switchboard.
However, that does not mean every Scope deployment across Solana is automatically independent of Switchboard.
The upstream sources used by an aggregator can change.
To determine whether a specific market depends on Switchboard, an auditor would need to inspect both:
- The market’s selected Scope account
- The source configuration behind the relevant Scope entry
Without that information, the true dependency may remain unknown.
A Stale Oracle Can Affect Protocols in Different Ways
The consequences of an outdated oracle depend on how a protocol handles stale data.
A lending protocol may reject transactions when a price becomes too old.
In that case, borrowing or other actions may temporarily fail.
A protocol that continues accepting stale data could instead face the risk of calculating positions using prices that no longer match the market.
Another system may automatically switch to a fallback provider.
Each outcome depends on the protocol’s freshness rules, fallback configuration, and live program settings.
Switchboard Pull Feeds Need Fresh Updates
Marginfi’s documentation says Switchboard pull feeds need to be refreshed before they are used.
That creates a different operational model from some other oracle systems.
Pyth push feeds, for example, are described as receiving updates through Pyth’s own infrastructure.
Scope uses aggregated values selected through configured account indexes.
Migrating between these systems can therefore change the accounts and transaction logic that applications must use.
Jito Has a Documented Fallback
Jito’s Tip Router documentation describes backup weights for situations where a price feed becomes unavailable.
If those fallback settings are still active, the system could potentially continue functioning even if Switchboard updates stop.
However, the current fallback values and live activation rules would need to be checked directly.
The existence of a fallback reduces the likelihood of an immediate total shutdown, but it does not prove that the system remains fully unaffected.
No Verified Switchboard Outage Has Been Shown
The article does not identify a confirmed outage at Jito, marginfi, or other major Solana protocols tied to the September 25 deadline.
There is also no verified evidence in the source showing forced liquidations or user losses caused by the Switchboard support shutdown.
A separate oracle incident on another platform earlier in September showed that incorrect prices can create serious consequences.
But that incident should not be treated as evidence that the same thing happened here.
Faster Solana Blocks Do Not Fix Oracle Problems
Solana’s faster block production also does not eliminate oracle risk.
A faster blockchain can publish fresh data quickly when that data exists.
But it cannot create a new price if the external oracle source is no longer updating.
Oracle freshness remains dependent on the provider and the protocol configuration rather than the speed of the underlying blockchain.
Protocols Are Responsible for Their Own Migration
The oracle provider supplies price data, but the consuming protocol decides which oracle account its program actually reads.
That means each protocol is ultimately responsible for migrating away from an unsupported feed.
Some markets may require governance action or administrator approval to change oracle addresses.
Front-end developers and third-party integrators may also need to update transaction logic to work with new account structures.
Users might not notice any of these technical changes unless a transaction fails or a market pauses.
Ending Support Does Not Mean Every Feed Instantly Stops
One of the most important distinctions is between ending official support and shutting down every live feed.
Switchboard’s announcement urged applications to migrate, but it did not establish a universal time when all feeds would stop functioning.
Some independent infrastructure may continue operating for a period of time.
Other feeds may become stale more quickly.
The only reliable way to determine the impact is to observe specific oracle accounts after the support deadline.
Fallbacks and Migrations Reduce Immediate Risk
Both marginfi and Jito show examples of how applications can reduce their dependency on a single oracle provider.
Marginfi introduced alternative oracle configurations before the deadline.
Jito documents a fallback system for unavailable pricing feeds.
These mechanisms suggest that a provider shutdown does not automatically create a system-wide failure.
However, neither example proves that every live market had successfully completed its migration by September 25.
A Market-by-Market Audit Is Still Needed
The largest missing piece is a live inventory of oracle configurations.
There is no publicly verified, synchronized list showing every active Solana market that still used Switchboard when support ended.
A proper audit would need to examine each market individually.
For every active lending bank, vault, or derivatives market, analysts would need to record:
- Program address
- Oracle type
- Oracle account
- Backup oracle
- Latest price update
- Maximum allowed price age
- Value of positions dependent on the oracle
Without those details, it is not possible to calculate how much capital was actually exposed.
Protocol TVL Is Not the Same as Oracle Exposure
It would also be misleading to take a protocol’s total value locked and assume all of it depends on Switchboard.
Different markets within the same protocol can use different pricing systems.
One market may use Pyth while another uses Scope or Switchboard.
Some oracle accounts may also remain in software even if no active borrowers or positions depend on them.
This means protocol-wide TVL cannot be used as a shortcut for measuring oracle risk.
What to Watch After Switchboard Support Ends
The most useful signals now are live technical indicators.
Analysts should monitor whether identified Switchboard feeds continue publishing fresh prices after September 25.
They should also check whether protocols announce migrations, pauses, or remaining dependencies.
Recent transaction history can show whether lending, reward distribution, and other operations continue normally.
Fallback systems should also be examined to understand what happens if a primary oracle becomes stale.
So Who Still Relies on Switchboard?
The most accurate answer is that public documentation identifies possible remaining dependencies, but it does not provide a verified live count.
Jito’s older documentation still shows Switchboard in its Tip Router pricing design and also describes a fallback mechanism.
Marginfi’s recent upgrade demonstrates clear technical paths away from Switchboard.
Kamino’s Scope architecture shows that an aggregator label alone cannot reveal the underlying oracle provider.
What remains missing is a market-by-market snapshot of live oracle configurations after the support deadline.
Until that information is available, claims about how many Solana applications or how much capital remains exposed to Switchboard should be treated cautiously.
The important distinction is simple: historical integration is not the same as current dependency.
