diff --git a/src/webgpu/shader/execution/expression/expression.ts b/src/webgpu/shader/execution/expression/expression.ts
index f85516f29bdd..0668005b4aeb 100644
--- a/src/webgpu/shader/execution/expression/expression.ts
+++ b/src/webgpu/shader/execution/expression/expression.ts
@@ -267,12 +267,12 @@ type PipelineCache = Map<String, GPUComputePipeline>;
  * @param create the function used to construct a value, if not found in the cache
  * @returns the value, either fetched from the cache, or newly built.
  */
-function getOrCreate<K, V>(map: Map<K, V>, key: K, create: () => V) {
+async function getOrCreate<K, V>(map: Map<K, V>, key: K, create: () => Promise<V>) {
   const existing = map.get(key);
   if (existing !== undefined) {
     return existing;
   }
-  const value = create();
+  const value = await create();
   map.set(key, value);
   return value;
 }
@@ -354,16 +354,24 @@ export async function run(
   };
 
   const processBatch = async (batchCases: CaseList) => {
-    const checkBatch = await submitBatch(
-      t,
-      shaderBuilder,
-      parameterTypes,
-      resultType,
-      batchCases,
-      cfg.inputSource,
-      pipelineCache
-    );
-    checkBatch();
+    try {
+      const checkBatch = await submitBatch(
+        t,
+        shaderBuilder,
+        parameterTypes,
+        resultType,
+        batchCases,
+        cfg.inputSource,
+        pipelineCache
+      );
+      checkBatch();
+    } catch (err) {
+      if (err instanceof GPUPipelineError) {
+        t.fail(`Pipeline Creation Error, ${err.reason}: ${err.message}`);
+      } else {
+        throw err;
+      }
+    }
     void t.queue.onSubmittedWorkDone().finally(batchFinishedCallback);
   };
 
@@ -1040,6 +1048,7 @@ async function buildPipeline(
       const module = t.device.createShaderModule({ code: source });
 
       // build the pipeline
+
       const pipeline = await t.device.createComputePipelineAsync({
         layout: 'auto',
         compute: { module, entryPoint: 'main' },
@@ -1084,12 +1093,12 @@ async function buildPipeline(
       }
 
       // build the compute pipeline, if the shader hasn't been compiled already.
-      const pipeline = getOrCreate(pipelineCache, source, () => {
+      const pipeline = await getOrCreate(pipelineCache, source, () => {
         // build the shader module
         const module = t.device.createShaderModule({ code: source });
 
         // build the pipeline
-        return t.device.createComputePipeline({
+        return t.device.createComputePipelineAsync({
           layout: 'auto',
           compute: { module, entryPoint: 'main' },
         });
diff --git a/src/webgpu/shader/execution/robust_access.spec.ts b/src/webgpu/shader/execution/robust_access.spec.ts
index 965dd283dd16..aafce2dcaae5 100644
--- a/src/webgpu/shader/execution/robust_access.spec.ts
+++ b/src/webgpu/shader/execution/robust_access.spec.ts
@@ -62,35 +62,43 @@ fn main() {
 
   t.debug(source);
   const module = t.device.createShaderModule({ code: source });
-  const pipeline = await t.device.createComputePipelineAsync({
-    layout,
-    compute: { module, entryPoint: 'main' },
-  });
-
-  const group = t.device.createBindGroup({
-    layout: pipeline.getBindGroupLayout(1),
-    entries: [
-      { binding: 0, resource: { buffer: constantsBuffer } },
-      { binding: 1, resource: { buffer: resultBuffer } },
-    ],
-  });
 
-  const testGroup = t.device.createBindGroup({
-    layout: pipeline.getBindGroupLayout(0),
-    entries: testBindings,
-  });
+  try {
+    const pipeline = await t.device.createComputePipelineAsync({
+      layout,
+      compute: { module, entryPoint: 'main' },
+    });
 
-  const encoder = t.device.createCommandEncoder();
-  const pass = encoder.beginComputePass();
-  pass.setPipeline(pipeline);
-  pass.setBindGroup(0, testGroup, dynamicOffsets);
-  pass.setBindGroup(1, group);
-  pass.dispatchWorkgroups(1);
-  pass.end();
+    const group = t.device.createBindGroup({
+      layout: pipeline.getBindGroupLayout(1),
+      entries: [
+        { binding: 0, resource: { buffer: constantsBuffer } },
+        { binding: 1, resource: { buffer: resultBuffer } },
+      ],
+    });
 
-  t.queue.submit([encoder.finish()]);
+    const testGroup = t.device.createBindGroup({
+      layout: pipeline.getBindGroupLayout(0),
+      entries: testBindings,
+    });
 
-  t.expectGPUBufferValuesEqual(resultBuffer, new Uint32Array([0]));
+    const encoder = t.device.createCommandEncoder();
+    const pass = encoder.beginComputePass();
+    pass.setPipeline(pipeline);
+    pass.setBindGroup(0, testGroup, dynamicOffsets);
+    pass.setBindGroup(1, group);
+    pass.dispatchWorkgroups(1);
+    pass.end();
+
+    t.queue.submit([encoder.finish()]);
+    t.expectGPUBufferValuesEqual(resultBuffer, new Uint32Array([0]));
+  } catch (err) {
+    if (err instanceof GPUPipelineError) {
+      t.fail(`Pipeline Creation Error, ${err.reason}: ${err.message}`);
+    } else {
+      throw err;
+    }
+  }
 }
 
 /** Fill an ArrayBuffer with sentinel values, except clear a region to zero. */
diff --git a/src/webgpu/shader/execution/zero_init.spec.ts b/src/webgpu/shader/execution/zero_init.spec.ts
index e03a72f8df56..eef155b68c62 100644
--- a/src/webgpu/shader/execution/zero_init.spec.ts
+++ b/src/webgpu/shader/execution/zero_init.spec.ts
@@ -446,101 +446,118 @@ g.test('compute,zero_init')
         ],
       });
 
-      const fillPipeline = await t.device.createComputePipelineAsync({
-        layout: t.device.createPipelineLayout({ bindGroupLayouts: [fillLayout] }),
-        label: 'Workgroup Fill Pipeline',
+      try {
+        const fillPipeline = await t.device.createComputePipelineAsync({
+          layout: t.device.createPipelineLayout({ bindGroupLayouts: [fillLayout] }),
+          label: 'Workgroup Fill Pipeline',
+          compute: {
+            module: t.device.createShaderModule({
+              code: wgsl,
+            }),
+            entryPoint: 'fill',
+          },
+        });
+
+        const inputBuffer = t.makeBufferWithContents(
+          new Uint32Array([...iterRange(wg_memory_limits / 4, _i => 0xdeadbeef)]),
+          GPUBufferUsage.STORAGE | GPUBufferUsage.COPY_DST
+        );
+        t.trackForCleanup(inputBuffer);
+        const outputBuffer = t.device.createBuffer({
+          size: wg_memory_limits,
+          usage: GPUBufferUsage.STORAGE | GPUBufferUsage.COPY_SRC,
+        });
+        t.trackForCleanup(outputBuffer);
+
+        const bg = t.device.createBindGroup({
+          layout: fillPipeline.getBindGroupLayout(0),
+          entries: [
+            {
+              binding: 0,
+              resource: {
+                buffer: inputBuffer,
+              },
+            },
+            {
+              binding: 1,
+              resource: {
+                buffer: outputBuffer,
+              },
+            },
+          ],
+        });
+
+        const e = t.device.createCommandEncoder();
+        const p = e.beginComputePass();
+        p.setPipeline(fillPipeline);
+        p.setBindGroup(0, bg);
+        p.dispatchWorkgroups(1);
+        p.end();
+        t.queue.submit([e.finish()]);
+      } catch (err) {
+        if (err instanceof GPUPipelineError) {
+          t.fail(`Pipeline Creation Error, ${err.reason}: ${err.message}`);
+          return;
+        } else {
+          throw err;
+        }
+      }
+    }
+
+    try {
+      const pipeline = await t.device.createComputePipelineAsync({
+        layout: 'auto',
         compute: {
           module: t.device.createShaderModule({
             code: wgsl,
           }),
-          entryPoint: 'fill',
+          entryPoint: 'main',
         },
       });
 
-      const inputBuffer = t.makeBufferWithContents(
-        new Uint32Array([...iterRange(wg_memory_limits / 4, _i => 0xdeadbeef)]),
-        GPUBufferUsage.STORAGE | GPUBufferUsage.COPY_DST
-      );
-      t.trackForCleanup(inputBuffer);
-      const outputBuffer = t.device.createBuffer({
-        size: wg_memory_limits,
+      const resultBuffer = t.device.createBuffer({
+        size: 4,
         usage: GPUBufferUsage.STORAGE | GPUBufferUsage.COPY_SRC,
       });
-      t.trackForCleanup(outputBuffer);
+      t.trackForCleanup(resultBuffer);
 
-      const bg = t.device.createBindGroup({
-        layout: fillPipeline.getBindGroupLayout(0),
+      const zeroBuffer = t.device.createBuffer({
+        size: 4,
+        usage: GPUBufferUsage.UNIFORM,
+      });
+      t.trackForCleanup(zeroBuffer);
+
+      const bindGroup = t.device.createBindGroup({
+        layout: pipeline.getBindGroupLayout(0),
         entries: [
           {
             binding: 0,
             resource: {
-              buffer: inputBuffer,
+              buffer: resultBuffer,
             },
           },
           {
             binding: 1,
             resource: {
-              buffer: outputBuffer,
+              buffer: zeroBuffer,
             },
           },
         ],
       });
 
-      const e = t.device.createCommandEncoder();
-      const p = e.beginComputePass();
-      p.setPipeline(fillPipeline);
-      p.setBindGroup(0, bg);
-      p.dispatchWorkgroups(1);
-      p.end();
-      t.queue.submit([e.finish()]);
+      const encoder = t.device.createCommandEncoder();
+      const pass = encoder.beginComputePass();
+      pass.setPipeline(pipeline);
+      pass.setBindGroup(0, bindGroup);
+      pass.dispatchWorkgroups(1);
+      pass.end();
+      t.queue.submit([encoder.finish()]);
+      t.expectGPUBufferValuesEqual(resultBuffer, new Uint32Array([0]));
+    } catch (err) {
+      if (err instanceof GPUPipelineError) {
+        t.fail(`Pipeline Creation Error, ${err.reason}: ${err.message}`);
+      } else {
+        throw err;
+      }
     }
-
-    const pipeline = await t.device.createComputePipelineAsync({
-      layout: 'auto',
-      compute: {
-        module: t.device.createShaderModule({
-          code: wgsl,
-        }),
-        entryPoint: 'main',
-      },
-    });
-
-    const resultBuffer = t.device.createBuffer({
-      size: 4,
-      usage: GPUBufferUsage.STORAGE | GPUBufferUsage.COPY_SRC,
-    });
-    t.trackForCleanup(resultBuffer);
-
-    const zeroBuffer = t.device.createBuffer({
-      size: 4,
-      usage: GPUBufferUsage.UNIFORM,
-    });
-    t.trackForCleanup(zeroBuffer);
-
-    const bindGroup = t.device.createBindGroup({
-      layout: pipeline.getBindGroupLayout(0),
-      entries: [
-        {
-          binding: 0,
-          resource: {
-            buffer: resultBuffer,
-          },
-        },
-        {
-          binding: 1,
-          resource: {
-            buffer: zeroBuffer,
-          },
-        },
-      ],
-    });
-
-    const encoder = t.device.createCommandEncoder();
-    const pass = encoder.beginComputePass();
-    pass.setPipeline(pipeline);
-    pass.setBindGroup(0, bindGroup);
-    pass.dispatchWorkgroups(1);
-    pass.end();
-    t.queue.submit([encoder.finish()]);
-    t.expectGPUBufferValuesEqual(resultBuffer, new Uint32Array([0]));
   });