diff --git a/http/zstd_decompressor.cpp b/http/zstd_decompressor.cpp
new file mode 100644
index 0000000..aa42392
--- /dev/null
+++ b/http/zstd_decompressor.cpp
@@ -0,0 +1,63 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright OpenBMC Authors
+
+#include "zstd_decompressor.hpp"
+
+#include "logging.hpp"
+
+#ifdef HAVE_ZSTD
+#include <zstd.h>
+#endif
+#include <boost/asio/buffer.hpp>
+
+#include <cstddef>
+#include <optional>
+
+#ifdef HAVE_ZSTD
+ZstdDecompressor::ZstdDecompressor() : dctx(ZSTD_createDStream())
+{
+    ZSTD_initDStream(dctx);
+}
+#else
+ZstdDecompressor::ZstdDecompressor() {};
+#endif
+
+std::optional<boost::asio::const_buffer> ZstdDecompressor::decompress(
+    [[maybe_unused]] boost::asio::const_buffer buffIn)
+{
+#ifdef HAVE_ZSTD
+    compressionBuf.clear();
+    ZSTD_inBuffer input = {buffIn.data(), buffIn.size(), 0};
+
+    // Note, this loop is prone to compression bombs, decompressing chunks that
+    // appear very small, but decompress to be very large, given that they're
+    // highly decompressible. This algorithm assumes that at this time, the
+    // whole file will fit in ram.
+    while (input.pos != input.size)
+    {
+        constexpr size_t frameSize = 4096;
+        auto buffer = compressionBuf.prepare(frameSize);
+        ZSTD_outBuffer output = {buffer.data(), buffer.size(), 0};
+        const size_t ret = ZSTD_decompressStream(dctx, &output, &input);
+        if (ZSTD_isError(ret) != 0)
+        {
+            BMCWEB_LOG_ERROR("Decompression Failed with code {}:{}", ret,
+                             ZSTD_getErrorName(ret));
+            return std::nullopt;
+        }
+        compressionBuf.commit(output.pos);
+    }
+    return compressionBuf.cdata();
+#else
+    BMCWEB_LOG_CRITICAL("Attempt to decompress, but libzstd not enabled");
+
+    return std::nullopt;
+#endif
+}
+
+ZstdDecompressor::~ZstdDecompressor()
+{
+#ifdef HAVE_ZSTD
+    ZSTD_freeDStream(dctx);
+#endif
+}