diff --git a/src/amberscript/parser.cc b/src/amberscript/parser.cc
index 72b3650e..ec8ad1e8 100644
--- a/src/amberscript/parser.cc
+++ b/src/amberscript/parser.cc
@@ -2952,7 +2952,8 @@ Result Parser::ParseBufferInitializerFill(Buffer* buffer,
   }
 
   auto fmt = buffer->GetFormat();
-  bool is_double_data = fmt->IsFloat32() || fmt->IsFloat64();
+  bool is_float_data =
+      fmt->IsFloat16() || fmt->IsFloat32() || fmt->IsFloat64();
 
   // Inflate the size because our items are multi-dimensional.
   size_in_items = size_in_items * fmt->InputNeededPerElement();
@@ -2960,7 +2961,7 @@ Result Parser::ParseBufferInitializerFill(Buffer* buffer,
   std::vector<Value> values;
   values.resize(size_in_items);
   for (size_t i = 0; i < size_in_items; ++i) {
-    if (is_double_data) {
+    if (is_float_data) {
       values[i].SetDoubleValue(token->AsDouble());
     } else {
       values[i].SetIntValue(token->AsUint64());
@@ -2994,7 +2995,8 @@ Result Parser::ParseBufferInitializerSeries(Buffer* buffer,
   auto n = type->AsNumber();
   FormatMode mode = n->GetFormatMode();
   uint32_t num_bits = n->NumBits();
-  if (type::Type::IsFloat32(mode, num_bits) ||
+  if (type::Type::IsFloat16(mode, num_bits) ||
+      type::Type::IsFloat32(mode, num_bits) ||
       type::Type::IsFloat64(mode, num_bits)) {
     counter.SetDoubleValue(token->AsDouble());
   } else {
@@ -3020,7 +3022,8 @@ Result Parser::ParseBufferInitializerSeries(Buffer* buffer,
   std::vector<Value> values;
   values.resize(size_in_items);
   for (size_t i = 0; i < size_in_items; ++i) {
-    if (type::Type::IsFloat32(mode, num_bits) ||
+    if (type::Type::IsFloat16(mode, num_bits) ||
+        type::Type::IsFloat32(mode, num_bits) ||
         type::Type::IsFloat64(mode, num_bits)) {
       double value = counter.AsDouble();
       values[i].SetDoubleValue(value);
diff --git a/src/amberscript/parser_buffer_test.cc b/src/amberscript/parser_buffer_test.cc
index ab4824ad..7ce18074 100644
--- a/src/amberscript/parser_buffer_test.cc
+++ b/src/amberscript/parser_buffer_test.cc
@@ -14,6 +14,7 @@
 
 #include "gtest/gtest.h"
 #include "src/amberscript/parser.h"
+#include "src/float16_helper.h"
 
 namespace amber {
 namespace amberscript {
@@ -400,6 +401,34 @@ TEST_F(AmberScriptParserTest, BufferFillFloat) {
   }
 }
 
+TEST_F(AmberScriptParserTest, BufferFillFloat16) {
+  std::string in = "BUFFER my_buffer DATA_TYPE float16 SIZE 5 FILL 5.5";
+
+  Parser parser;
+  Result r = parser.Parse(in);
+  ASSERT_TRUE(r.IsSuccess()) << r.Error();
+
+  auto script = parser.GetScript();
+  const auto& buffers = script->GetBuffers();
+  ASSERT_EQ(1U, buffers.size());
+
+  ASSERT_TRUE(buffers[0] != nullptr);
+
+  auto* buffer = buffers[0].get();
+  EXPECT_EQ("my_buffer", buffer->GetName());
+  EXPECT_EQ(Format::Layout::kStd430, buffer->GetFormat()->GetLayout());
+  EXPECT_EQ(5U, buffer->ElementCount());
+  EXPECT_EQ(5U, buffer->ValueCount());
+  EXPECT_EQ(5U * sizeof(uint16_t), buffer->GetSizeInBytes());
+
+  const auto* data = buffer->GetValues<uint16_t>();
+  for (size_t i = 0; i < buffer->ValueCount(); ++i) {
+    EXPECT_FLOAT_EQ(
+        5.5f, float16::HexFloatToFloat(
+                  reinterpret_cast<const uint8_t*>(&data[i]), 16));
+  }
+}
+
 TEST_F(AmberScriptParserTest, BufferSeries) {
   std::string in =
       "BUFFER my_buffer DATA_TYPE uint8 SIZE 5 SERIES_FROM 2 INC_BY 1";
@@ -461,6 +490,38 @@ TEST_F(AmberScriptParserTest, BufferSeriesFloat) {
   }
 }
 
+TEST_F(AmberScriptParserTest, BufferSeriesFloat16) {
+  std::string in =
+      "BUFFER my_buffer DATA_TYPE float16 SIZE 5 SERIES_FROM 2.5 INC_BY "
+      "0.5";
+
+  Parser parser;
+  Result r = parser.Parse(in);
+  ASSERT_TRUE(r.IsSuccess()) << r.Error();
+
+  auto script = parser.GetScript();
+  const auto& buffers = script->GetBuffers();
+  ASSERT_EQ(1U, buffers.size());
+
+  ASSERT_TRUE(buffers[0] != nullptr);
+
+  auto* buffer = buffers[0].get();
+  EXPECT_EQ("my_buffer", buffer->GetName());
+  EXPECT_EQ(Format::Layout::kStd430, buffer->GetFormat()->GetLayout());
+  EXPECT_EQ(5U, buffer->ElementCount());
+  EXPECT_EQ(5U, buffer->ValueCount());
+  EXPECT_EQ(5U * sizeof(uint16_t), buffer->GetSizeInBytes());
+
+  std::vector<float> results = {2.5f, 3.0f, 3.5f, 4.0f, 4.5f};
+  const auto* data = buffer->GetValues<uint16_t>();
+  ASSERT_EQ(results.size(), buffer->ValueCount());
+  for (size_t i = 0; i < results.size(); ++i) {
+    EXPECT_FLOAT_EQ(
+        results[i], float16::HexFloatToFloat(
+                        reinterpret_cast<const uint8_t*>(&data[i]), 16));
+  }
+}
+
 TEST_F(AmberScriptParserTest, BufferMultipleBuffers) {
   std::string in = R"(
 BUFFER color_buffer DATA_TYPE uint8 SIZE 5 FILL 5
diff --git a/src/format.h b/src/format.h
index 2e917058..4eac62bb 100644
--- a/src/format.h
+++ b/src/format.h
@@ -199,6 +199,12 @@ class Format {
            type::Type::IsUint64(type_->AsNumber()->GetFormatMode(),
                                 type_->AsNumber()->NumBits());
   }
+  /// Returns true if all components of this format are a 16 bit float.
+  bool IsFloat16() const {
+    return type_->IsNumber() &&
+           type::Type::IsFloat16(type_->AsNumber()->GetFormatMode(),
+                                 type_->AsNumber()->NumBits());
+  }
   /// Returns true if all components of this format are a 32 bit float.
   bool IsFloat32() const {
     return type_->IsNumber() &&