diff --git a/nexus/db-queries/src/authn/mod.rs b/nexus/db-queries/src/authn/mod.rs
index 76824c7d081..305c359820a 100644
--- a/nexus/db-queries/src/authn/mod.rs
+++ b/nexus/db-queries/src/authn/mod.rs
@@ -228,8 +228,8 @@ impl Context {
         )
     }
 
-    /// Returns an authenticated context for the specific Silo user.
-    #[cfg(test)]
+    /// Returns an authenticated context for the specific Silo user. Not marked
+    /// as #[cfg(test)] so that this is available in integration tests.
     pub fn for_test_user(
         silo_user_id: Uuid,
         silo_id: Uuid,
diff --git a/nexus/src/app/sagas/instance_start.rs b/nexus/src/app/sagas/instance_start.rs
index 5d02d44b6b3..068d2e70055 100644
--- a/nexus/src/app/sagas/instance_start.rs
+++ b/nexus/src/app/sagas/instance_start.rs
@@ -323,6 +323,8 @@ async fn sis_dpd_ensure(
     );
     let datastore = osagactx.datastore();
 
+    // Querying sleds requires fleet access; use the instance allocator context
+    // for this.
     let sled_uuid = sagactx.lookup::<Uuid>("sled_id")?;
     let (.., sled) = LookupPath::new(&osagactx.nexus().opctx_alloc, &datastore)
         .sled_id(sled_uuid)
@@ -330,9 +332,11 @@ async fn sis_dpd_ensure(
         .await
         .map_err(ActionError::action_failed)?;
 
+    // Querying boundary switches also requires fleet access and the use of the
+    // instance allocator context.
     let boundary_switches = osagactx
         .nexus()
-        .boundary_switches(&opctx)
+        .boundary_switches(&osagactx.nexus().opctx_alloc)
         .await
         .map_err(ActionError::action_failed)?;
 
diff --git a/nexus/src/app/test_interfaces.rs b/nexus/src/app/test_interfaces.rs
index 486569333ea..c7a6165998c 100644
--- a/nexus/src/app/test_interfaces.rs
+++ b/nexus/src/app/test_interfaces.rs
@@ -24,6 +24,12 @@ pub trait TestInterfaces {
         id: &Uuid,
     ) -> Result<Option<Arc<SledAgentClient>>, Error>;
 
+    async fn instance_sled_by_id_with_opctx(
+        &self,
+        id: &Uuid,
+        opctx: &OpContext,
+    ) -> Result<Option<Arc<SledAgentClient>>, Error>;
+
     /// Returns the SledAgentClient for the sled running an instance to which a
     /// disk is attached.
     async fn disk_sled_by_id(
@@ -37,6 +43,12 @@ pub trait TestInterfaces {
         instance_id: &Uuid,
     ) -> Result<Option<Uuid>, Error>;
 
+    async fn instance_sled_id_with_opctx(
+        &self,
+        instance_id: &Uuid,
+        opctx: &OpContext,
+    ) -> Result<Option<Uuid>, Error>;
+
     async fn set_disk_as_faulted(&self, disk_id: &Uuid) -> Result<bool, Error>;
 
     fn set_samael_max_issue_delay(&self, max_issue_delay: chrono::Duration);
@@ -52,7 +64,20 @@ impl TestInterfaces for super::Nexus {
         &self,
         id: &Uuid,
     ) -> Result<Option<Arc<SledAgentClient>>, Error> {
-        let sled_id = self.instance_sled_id(id).await?;
+        let opctx = OpContext::for_tests(
+            self.log.new(o!()),
+            Arc::clone(&self.db_datastore),
+        );
+
+        self.instance_sled_by_id_with_opctx(id, &opctx).await
+    }
+
+    async fn instance_sled_by_id_with_opctx(
+        &self,
+        id: &Uuid,
+        opctx: &OpContext,
+    ) -> Result<Option<Arc<SledAgentClient>>, Error> {
+        let sled_id = self.instance_sled_id_with_opctx(id, opctx).await?;
         if let Some(sled_id) = sled_id {
             Ok(Some(self.sled_client(&sled_id).await?))
         } else {
@@ -83,6 +108,14 @@ impl TestInterfaces for super::Nexus {
             Arc::clone(&self.db_datastore),
         );
 
+        self.instance_sled_id_with_opctx(id, &opctx).await
+    }
+
+    async fn instance_sled_id_with_opctx(
+        &self,
+        id: &Uuid,
+        opctx: &OpContext,
+    ) -> Result<Option<Uuid>, Error> {
         let (.., authz_instance) = LookupPath::new(&opctx, &self.db_datastore)
             .instance_id(*id)
             .lookup_for(nexus_db_queries::authz::Action::Read)
@@ -90,7 +123,7 @@ impl TestInterfaces for super::Nexus {
 
         Ok(self
             .datastore()
-            .instance_fetch_with_vmm(&opctx, &authz_instance)
+            .instance_fetch_with_vmm(opctx, &authz_instance)
             .await?
             .sled_id())
     }
diff --git a/nexus/tests/integration_tests/instances.rs b/nexus/tests/integration_tests/instances.rs
index b8fcc9f2cb1..9208e216521 100644
--- a/nexus/tests/integration_tests/instances.rs
+++ b/nexus/tests/integration_tests/instances.rs
@@ -3611,10 +3611,11 @@ async fn test_instance_create_in_silo(cptestctx: &ControlPlaneTestContext) {
 
     // Create an instance using the authorization granted to the collaborator
     // Silo User.
+    let instance_name = "collaborate-with-me";
     let instance_params = params::InstanceCreate {
         identity: IdentityMetadataCreateParams {
-            name: Name::try_from(String::from("ip-pool-test")).unwrap(),
-            description: String::from("instance to test IP Pool authz"),
+            name: Name::try_from(String::from(instance_name)).unwrap(),
+            description: String::from("instance to test creation in a silo"),
         },
         ncpus: InstanceCpuCount::try_from(2).unwrap(),
         memory: ByteCount::from_gibibytes_u32(4),
@@ -3635,6 +3636,34 @@ async fn test_instance_create_in_silo(cptestctx: &ControlPlaneTestContext) {
         .expect("Failed to create instance")
         .parsed_body::<Instance>()
         .expect("Failed to parse instance");
+
+    // Make sure the instance can actually start even though a collaborator
+    // created it.
+    let apictx = &cptestctx.server.apictx();
+    let nexus = &apictx.nexus;
+    let authn = AuthnMode::SiloUser(user_id);
+    let instance_url = get_instance_url(instance_name);
+    let instance = instance_get_as(&client, &instance_url, authn.clone()).await;
+    info!(&cptestctx.logctx.log, "test got instance"; "instance" => ?instance);
+    assert_eq!(instance.runtime.run_state, InstanceState::Starting);
+
+    // The default instance simulation function uses a test user that, while
+    // privileged, only has access to the default silo. Synthesize an operation
+    // context that grants access to the silo under test.
+    let opctx = OpContext::for_background(
+        cptestctx.logctx.log.new(o!()),
+        Arc::new(nexus_db_queries::authz::Authz::new(&cptestctx.logctx.log)),
+        nexus_db_queries::authn::Context::for_test_user(
+            user_id,
+            silo.identity.id,
+            nexus_db_queries::authn::SiloAuthnPolicy::try_from(&*DEFAULT_SILO)
+                .unwrap(),
+        ),
+        nexus.datastore().clone(),
+    );
+    instance_simulate_with_opctx(nexus, &instance.identity.id, &opctx).await;
+    let instance = instance_get_as(&client, &instance_url, authn).await;
+    assert_eq!(instance.runtime.run_state, InstanceState::Running);
 }
 
 /// Test that appropriate OPTE V2P mappings are created and deleted.
@@ -3752,9 +3781,17 @@ async fn test_instance_v2p_mappings(cptestctx: &ControlPlaneTestContext) {
 async fn instance_get(
     client: &ClientTestContext,
     instance_url: &str,
+) -> Instance {
+    instance_get_as(client, instance_url, AuthnMode::PrivilegedUser).await
+}
+
+async fn instance_get_as(
+    client: &ClientTestContext,
+    instance_url: &str,
+    authn_as: AuthnMode,
 ) -> Instance {
     NexusRequest::object_get(client, instance_url)
-        .authn_as(AuthnMode::PrivilegedUser)
+        .authn_as(authn_as)
         .execute()
         .await
         .unwrap()
@@ -3857,6 +3894,19 @@ pub async fn instance_simulate(nexus: &Arc<Nexus>, id: &Uuid) {
     sa.instance_finish_transition(*id).await;
 }
 
+pub async fn instance_simulate_with_opctx(
+    nexus: &Arc<Nexus>,
+    id: &Uuid,
+    opctx: &OpContext,
+) {
+    let sa = nexus
+        .instance_sled_by_id_with_opctx(id, opctx)
+        .await
+        .unwrap()
+        .expect("instance must be on a sled to simulate a state change");
+    sa.instance_finish_transition(*id).await;
+}
+
 /// Simulates state transitions for the incarnation of the instance on the
 /// supplied sled (which may not be the sled ID currently stored in the
 /// instance's CRDB record).