diff --git a/compilers/concrete-compiler/compiler/lib/Conversion/FHEToTFHECrt/FHEToTFHECrt.cpp b/compilers/concrete-compiler/compiler/lib/Conversion/FHEToTFHECrt/FHEToTFHECrt.cpp
index abef451c7..3bcb127f0 100644
--- a/compilers/concrete-compiler/compiler/lib/Conversion/FHEToTFHECrt/FHEToTFHECrt.cpp
+++ b/compilers/concrete-compiler/compiler/lib/Conversion/FHEToTFHECrt/FHEToTFHECrt.cpp
@@ -915,6 +915,29 @@ struct InsertSliceOpPattern : public CrtOpPattern<mlir::tensor::InsertSliceOp> {
   };
 };
 
+/// Zero op result can be a tensor after CRT encoding, and thus need to be
+/// rewritten as a ZeroTensor op
+struct ZeroOpPattern : public CrtOpPattern<FHE::ZeroEintOp> {
+  ZeroOpPattern(mlir::MLIRContext *context,
+                mlir::concretelang::CrtLoweringParameters params,
+                mlir::PatternBenefit benefit = 1)
+      : CrtOpPattern<FHE::ZeroEintOp>(context, params, benefit) {}
+
+  ::mlir::LogicalResult
+  matchAndRewrite(FHE::ZeroEintOp op, FHE::ZeroEintOp::Adaptor adaptor,
+                  ::mlir::ConversionPatternRewriter &rewriter) const override {
+
+    mlir::TypeConverter *converter = this->getTypeConverter();
+    auto glweOrTensorType = converter->convertType(op.getResult().getType());
+    if (mlir::dyn_cast<mlir::TensorType>(glweOrTensorType)) {
+      rewriter.replaceOpWithNewOp<TFHE::ZeroTensorGLWEOp>(op, glweOrTensorType);
+    } else {
+      rewriter.replaceOpWithNewOp<TFHE::ZeroGLWEOp>(op, glweOrTensorType);
+    }
+    return mlir::success();
+  };
+};
+
 } // namespace lowering
 
 struct FHEToTFHECrtPass : public FHEToTFHECrtBase<FHEToTFHECrtPass> {
@@ -983,10 +1006,8 @@ struct FHEToTFHECrtPass : public FHEToTFHECrtBase<FHEToTFHECrtPass> {
                                                    converter);
 
     // Patterns for the `FHE` dialect operations
+    patterns.add<lowering::ZeroOpPattern>(&getContext(), loweringParameters);
     patterns.add<
-        //    |_ `FHE::zero_eint`
-        mlir::concretelang::GenericOneToOneOpConversionPattern<
-            FHE::ZeroEintOp, TFHE::ZeroGLWEOp>,
         //    |_ `FHE::zero_tensor`
         mlir::concretelang::GenericOneToOneOpConversionPattern<
             FHE::ZeroTensorOp, TFHE::ZeroTensorGLWEOp>>(&getContext(),