diff --git a/mlir/include/mlir/Dialect/Linalg/Transforms/Transforms.h b/mlir/include/mlir/Dialect/Linalg/Transforms/Transforms.h
--- a/mlir/include/mlir/Dialect/Linalg/Transforms/Transforms.h
+++ b/mlir/include/mlir/Dialect/Linalg/Transforms/Transforms.h
@@ -158,8 +158,8 @@
   SmallVector<Operation *, 8> loops;
   SmallVector<Value, 4> tensorResults;
 };
-Optional<TiledLinalgOp> tileLinalgOp(OpBuilder &b, LinalgOp op,
-                                     const LinalgTilingOptions &options);
+FailureOr<TiledLinalgOp> tileLinalgOp(OpBuilder &b, LinalgOp op,
+                                      const LinalgTilingOptions &options);
 
 /// Fuse a sequence of linalg operations (`ops`) using tile-and-fuse. This
 /// proceeds as follows:
diff --git a/mlir/lib/Dialect/Linalg/Transforms/Tiling.cpp b/mlir/lib/Dialect/Linalg/Transforms/Tiling.cpp
--- a/mlir/lib/Dialect/Linalg/Transforms/Tiling.cpp
+++ b/mlir/lib/Dialect/Linalg/Transforms/Tiling.cpp
@@ -152,7 +152,7 @@
 }
 
 template <typename LoopTy>
-static Optional<TiledLinalgOp>
+static FailureOr<TiledLinalgOp>
 tileLinalgOpImpl(OpBuilder &b, LinalgOp op, ValueRange tileSizes,
                  const LinalgTilingOptions &options) {
   auto nLoops = op.getNumLoops();
@@ -160,13 +160,13 @@
   tileSizes = tileSizes.take_front(nLoops);
 
   if (llvm::all_of(tileSizes, isZero))
-    return llvm::None;
+    return failure();
 
   // 1. Build the tiled loop ranges.
   auto allShapeSizes = op.createFlatListOfOperandDims(b, op.getLoc());
   AffineMap shapeSizesToLoopsMap = op.getShapesToLoopsMap();
   if (!shapeSizesToLoopsMap)
-    return llvm::None;
+    return failure();
 
   SmallVector<Range, 4> loopRanges;
   LoopIndexToRangeIndexMap loopIndexToRangeIndex;
@@ -291,13 +291,13 @@
 }
 
 template <typename LoopTy>
-Optional<TiledLinalgOp> static tileLinalgOpImpl(
+FailureOr<TiledLinalgOp> static tileLinalgOpImpl(
     OpBuilder &b, LinalgOp op, const LinalgTilingOptions &options) {
   OpBuilder::InsertionGuard g(b);
   b.setInsertionPoint(op);
 
   if (!options.tileSizeComputationFunction)
-    return llvm::None;
+    return failure();
 
   // Enforce the convention that "tiling by zero" skips tiling a particular
   // dimension. This convention is significantly simpler to handle instead of
@@ -313,7 +313,7 @@
   return tileLinalgOpImpl<LoopTy>(b, op, tileSizeVector, options);
 }
 
-Optional<TiledLinalgOp>
+FailureOr<TiledLinalgOp>
 mlir::linalg::tileLinalgOp(OpBuilder &b, LinalgOp op,
                            const LinalgTilingOptions &options) {
   switch (options.loopType) {
@@ -325,7 +325,7 @@
     return tileLinalgOpImpl<linalg::TiledLoopOp>(b, op, options);
   default:;
   }
-  return llvm::None;
+  return failure();
 }
 
 /// Generate a loop nest around a given PadTensorOp (for tiling). `newPadOp`