diff --git a/headroom/dashboard/templates/dashboard.html b/headroom/dashboard/templates/dashboard.html
index 67a696f15..a4c9988f5 100644
--- a/headroom/dashboard/templates/dashboard.html
+++ b/headroom/dashboard/templates/dashboard.html
@@ -105,12 +105,12 @@
             <div class="bg-surface rounded-lg p-4 border border-border">
                 <div class="text-xs text-gray-500 uppercase tracking-wide mb-1">Proxy $ Saved</div>
                 <div class="flex items-baseline gap-2">
-                    <span class="text-3xl font-light tabular-nums text-emerald-400" x-text="'$' + formatCurrency(stats.cost?.savings_usd || 0)"></span>
+                    <span class="text-3xl font-light tabular-nums text-emerald-400" x-text="'$' + formatCurrency(stats.persistent_savings?.lifetime?.compression_savings_usd || 0)"></span>
                 </div>
                 <div class="mt-2 text-xs text-gray-500">
-                    <span x-show="stats.cost?.savings_usd > 0"
+                    <span x-show="(stats.persistent_savings?.lifetime?.compression_savings_usd || 0) > 0"
                           x-text="formatNumber(stats.tokens?.proxy_compression_saved || 0) + ' proxy tokens only; ' + cliFilteringLabel + ' excluded from $'"></span>
-                    <span x-show="!(stats.cost?.savings_usd > 0)"
+                    <span x-show="!((stats.persistent_savings?.lifetime?.compression_savings_usd || 0) > 0)"
                           x-text="formatNumber(stats.requests?.total || 0) + ' requests processed'"></span>
                 </div>
             </div>
diff --git a/headroom/proxy/server.py b/headroom/proxy/server.py
index f8b68a85a..5bb22ba84 100644
--- a/headroom/proxy/server.py
+++ b/headroom/proxy/server.py
@@ -3099,12 +3099,13 @@ def run_server(
         # sticky-session workaround.
         logger.warning(
             "Headroom is running with workers=%d. The in-memory CCR store, "
-            "compression cache, prefix tracker, and TOIN state are all "
+            "compression cache, prefix tracker, TOIN state, and CostTracker are all "
             "per-process; multi-worker deployments produce silent retrieval "
-            "failures and avoidable cache busts when sessions land on different "
-            "workers. Run --workers 1 (or place a sticky-session load balancer "
-            "in front of multiple --workers 1 processes). See RUST_DEV.md → "
-            "'Multi-worker deployment — CCR fragmentation'.",
+            "failures, avoidable cache busts, and an unstable dashboard 'Proxy $ Saved' "
+            "hero tile (each /stats poll hits a different worker's partial total) when "
+            "sessions land on different workers. Run --workers 1 (or place a "
+            "sticky-session load balancer in front of multiple --workers 1 processes). "
+            "See RUST_DEV.md → 'Multi-worker deployment — CCR fragmentation'.",
             workers,
         )
         os.environ[_MULTI_WORKER_CONFIG_ENV] = json.dumps(_proxy_config_payload(config))
diff --git a/tests/test_proxy_savings_history.py b/tests/test_proxy_savings_history.py
index e4874f5a6..8e95c05f4 100644
--- a/tests/test_proxy_savings_history.py
+++ b/tests/test_proxy_savings_history.py
@@ -211,6 +211,42 @@ def test_record_compression_savings_skips_empty_updates_and_normalizes_timestamp
     assert persisted["history"][-1]["timestamp"] == "2026-03-27T12:34:00Z"
 
 
+def test_savings_tracker_save_does_not_flock_target_inode_before_replace(tmp_path, monkeypatch):
+    path = tmp_path / "proxy_savings.json"
+    tracker = SavingsTracker(path=str(path))
+
+    tracker.record_request(
+        model="gpt-4o",
+        input_tokens=120,
+        tokens_saved=10,
+        timestamp="2026-03-27T09:00:00Z",
+    )
+    assert path.exists()
+
+    flock_calls: list[int] = []
+
+    class _FcntlSpy:
+        LOCK_EX = 1
+        LOCK_UN = 2
+
+        def flock(self, _fh, operation: int) -> None:
+            flock_calls.append(operation)
+
+    monkeypatch.setattr(savings_tracker_module, "_HAS_FCNTL", True, raising=False)
+    monkeypatch.setattr(savings_tracker_module, "_fcntl", _FcntlSpy(), raising=False)
+
+    tracker.record_request(
+        model="gpt-4o",
+        input_tokens=80,
+        tokens_saved=5,
+        timestamp="2026-03-27T09:10:00Z",
+    )
+
+    assert flock_calls == []
+    persisted = json.loads(path.read_text(encoding="utf-8"))
+    assert persisted["lifetime"]["tokens_saved"] == 15
+
+
 def test_litellm_resolution_and_savings_estimation_fallbacks(monkeypatch):
     def fake_cost_per_token(*, model, prompt_tokens, completion_tokens):
         if model in {"gpt-4o", "anthropic/claude-sonnet-4-6"}: