diff --git a/src/inference/src/dev/threading/cpu_streams_executor.cpp b/src/inference/src/dev/threading/cpu_streams_executor.cpp
index 0b22e913628..2a45314f0ed 100644
--- a/src/inference/src/dev/threading/cpu_streams_executor.cpp
+++ b/src/inference/src/dev/threading/cpu_streams_executor.cpp
@@ -133,9 +133,7 @@ struct CPUStreamsExecutor::Impl {
                                    const int concurrency,
                                    const int core_type,
                                    const int numa_node_id) {
-            _numaNodeId = (_impl->_usedNumaNodes.size() == 1 && _impl->_usedNumaNodes.at(0) == -1)
-                              ? -1  // macOS
-                              : std::max(0, numa_node_id);
+            _numaNodeId = std::max(0, numa_node_id);
             _socketId = get_socket_by_numa_node(_numaNodeId);
             if (stream_type == STREAM_WITHOUT_PARAM) {
                 _taskArena.reset(new custom::task_arena{concurrency});
diff --git a/src/inference/src/system_conf.cpp b/src/inference/src/system_conf.cpp
index 16fe74bf754..fcd018aa467 100644
--- a/src/inference/src/system_conf.cpp
+++ b/src/inference/src/system_conf.cpp
@@ -247,7 +247,7 @@ int get_num_numa_nodes() {
     return cpu_info()._numa_nodes;
 }
 int get_num_sockets() {
-    return -1;
+    return cpu_info()._sockets;
 }
 void reserve_available_cpus(const std::vector<std::vector<int>> streams_info_table,
                             std::vector<std::vector<int>>& stream_processors,
@@ -255,6 +255,12 @@ void reserve_available_cpus(const std::vector<std::vector<int>> streams_info_tab
 void set_cpu_used(const std::vector<int>& cpu_ids, const int used) {}
 
 int get_socket_by_numa_node(int numa_node_id) {
+    CPU& cpu = cpu_info();
+    for (size_t i = 0; i < cpu._proc_type_table.size(); i++) {
+        if (cpu._proc_type_table[i][PROC_NUMA_NODE_ID] == numa_node_id) {
+            return cpu._proc_type_table[i][PROC_SOCKET_ID];
+        }
+    }
     return -1;
 };
 
diff --git a/src/plugins/intel_cpu/src/exec_network.cpp b/src/plugins/intel_cpu/src/exec_network.cpp
index 21ac20a2942..b779891df09 100644
--- a/src/plugins/intel_cpu/src/exec_network.cpp
+++ b/src/plugins/intel_cpu/src/exec_network.cpp
@@ -176,9 +176,7 @@ ExecNetwork::GraphGuard::Lock ExecNetwork::GetGraph() const {
                 {
                     std::lock_guard<std::mutex> lock{*_mutex.get()};
                     // disable weights caching if graph was created only once
-                    // "socketId != -1" is the WA for MacOS, will remove later
-                    auto weightsCache =
-                        (_cfg.streamExecutorConfig._streams != 1 && socketId != -1) ? _socketWeights[socketId] : nullptr;
+                    auto weightsCache = _cfg.streamExecutorConfig._streams != 1 ? _socketWeights[socketId] : nullptr;
 
                     auto isQuantizedFlag =
                         (_cfg.lpTransformsMode == Config::On) &&