diff --git a/packages/rendermime/src/renderers.ts b/packages/rendermime/src/renderers.ts
index 65c2a31331b5..b608c723fdd0 100644
--- a/packages/rendermime/src/renderers.ts
+++ b/packages/rendermime/src/renderers.ts
@@ -796,6 +796,19 @@ export function renderText(options: renderText.IRenderOptions): Promise<void> {
   return Promise.resolve(undefined);
 }
 
+/**
+ * Sanitize HTML out using native browser sanitizer.
+ *
+ * Compared to the `ISanitizer.sanitize` this does not allow to selectively
+ * allow to keep certain tags but escapes everything; on the other hand
+ * it is much faster as it uses platform-optimized code.
+ */
+function nativeSanitize(source: string): string {
+  const el = document.createElement('span');
+  el.textContent = source;
+  return el.innerHTML;
+}
+
 /**
  * Render the textual representation into a host node.
  *
@@ -808,10 +821,16 @@ function renderTextual(
   // Unpack the options.
   const { host, sanitizer, source } = options;
 
-  // Create the HTML content.
-  const content = sanitizer.sanitize(Private.ansiSpan(source), {
-    allowedTags: ['span']
-  });
+  const ansiPrefixRe = /\x1b/; // eslint-disable-line no-control-regex
+  const hasAnsiPrefix = ansiPrefixRe.test(source);
+
+  // Create the HTML content:
+  // If no ANSI codes are present use a fast path for escaping.
+  const content = hasAnsiPrefix
+    ? sanitizer.sanitize(Private.ansiSpan(source), {
+        allowedTags: ['span']
+      })
+    : nativeSanitize(source);
 
   // Set the sanitized content for the host node.
   const pre = document.createElement('pre');
diff --git a/packages/rendermime/test/factories.spec.ts b/packages/rendermime/test/factories.spec.ts
index d9d3db2463e2..0cb732a2578c 100644
--- a/packages/rendermime/test/factories.spec.ts
+++ b/packages/rendermime/test/factories.spec.ts
@@ -630,6 +630,48 @@ describe('rendermime/factories', () => {
         expect(end - start).toBeLessThan(timeout);
       });
 
+      it('should use a fast path when no ANSI codes are present', async () => {
+        const mimeType = 'application/vnd.jupyter.stderr';
+
+        const ansiEscape = '\x1b[01;41;32mtext\x1b[00m';
+        const notAnsiEscape = '\x1a[01;41;32mtext\x1a[00m';
+
+        // We cannot just compare times here because:
+        // a) tests are run in jsdom thus "native" sanitizer is not much faster
+        // b) `Private.ansiSpan` has much higher cost when ANSI escapes are present
+
+        const testSource = '<script>window.x = 1</script>';
+        const spy = jest.spyOn(sanitizer, 'sanitize');
+
+        // Initialize slow path scenario
+        let model = createModel(mimeType, testSource + ansiEscape);
+        let w = errorRendererFactory.createRenderer({ mimeType, ...options });
+        expect(spy).toHaveBeenCalledTimes(0);
+
+        // Test slow path
+        await w.renderModel(model);
+        // Sanitizer.sanitize should have been called
+        expect(spy).toHaveBeenCalledTimes(1);
+
+        const escapedSlow = w.node.querySelector('pre')!.innerHTML;
+
+        // Initialize fast path scenario.
+        model = createModel(mimeType, testSource + notAnsiEscape);
+        w = errorRendererFactory.createRenderer({ mimeType, ...options });
+
+        // Test fast path
+        await w.renderModel(model);
+        // Sanitizer.sanitize should not have been called
+        expect(spy).toHaveBeenCalledTimes(1);
+
+        const escapedFast = w.node.querySelector('pre')!.innerHTML;
+
+        // Disregarding the suffix the escaped code should be the same.
+        expect(escapedFast.slice(0, testSource.length)).toEqual(
+          escapedSlow.slice(0, testSource.length)
+        );
+      });
+
       it.each([
         ['arrives in a new line', 'www.example.com', '\n a new line of text'],
         ['arrives after a new line', 'www.example.com\n', 'a new line of text'],