diff --git a/plat/ti/k3/board/generic/board.mk b/plat/ti/k3/board/generic/board.mk
index c5bd1ec3c..ef74cd64c 100644
--- a/plat/ti/k3/board/generic/board.mk
+++ b/plat/ti/k3/board/generic/board.mk
@@ -13,6 +13,10 @@ $(eval $(call add_define,PRELOADED_BL33_BASE))
 K3_HW_CONFIG_BASE ?= 0x82000000
 $(eval $(call add_define,K3_HW_CONFIG_BASE))
 
+# Define sec_proxy usage as the full prioritized communication scheme
+K3_SEC_PROXY_LITE	:=	0
+$(eval $(call add_define,K3_SEC_PROXY_LITE))
+
 # System coherency is managed in hardware
 USE_COHERENT_MEM	:=	1
 
diff --git a/plat/ti/k3/common/drivers/sec_proxy/sec_proxy.c b/plat/ti/k3/common/drivers/sec_proxy/sec_proxy.c
index cbe5fdab9..a0bfdee36 100644
--- a/plat/ti/k3/common/drivers/sec_proxy/sec_proxy.c
+++ b/plat/ti/k3/common/drivers/sec_proxy/sec_proxy.c
@@ -97,11 +97,16 @@ static struct k3_sec_proxy_mbox spm = {
 		.data_end_offset = 0x3C,
 	},
 	.threads = {
+#if !K3_SEC_PROXY_LITE
 		SP_THREAD(SP_NOTIFY),
 		SP_THREAD(SP_RESPONSE),
 		SP_THREAD(SP_HIGH_PRIORITY),
 		SP_THREAD(SP_LOW_PRIORITY),
 		SP_THREAD(SP_NOTIFY_RESP),
+#else
+		SP_THREAD(SP_RESPONSE),
+		SP_THREAD(SP_HIGH_PRIORITY),
+#endif /* K3_SEC_PROXY_LITE */
 	},
 };
 
diff --git a/plat/ti/k3/common/drivers/sec_proxy/sec_proxy.h b/plat/ti/k3/common/drivers/sec_proxy/sec_proxy.h
index 6c4f5dfff..f4b0b4bac 100644
--- a/plat/ti/k3/common/drivers/sec_proxy/sec_proxy.h
+++ b/plat/ti/k3/common/drivers/sec_proxy/sec_proxy.h
@@ -16,13 +16,28 @@
  * enum k3_sec_proxy_chan_id - Secure Proxy thread IDs
  *
  * These the available IDs used in k3_sec_proxy_{send,recv}()
+ * There are two schemes we use:
+ * * if K3_SEC_PROXY_LITE = 1, we just have two threads to talk
+ * * if K3_SEC_PROXY_LITE = 0, we have the full fledged
+ *   communication scheme available.
  */
 enum k3_sec_proxy_chan_id {
+#if !K3_SEC_PROXY_LITE
 	SP_NOTIFY = 0,
 	SP_RESPONSE,
 	SP_HIGH_PRIORITY,
 	SP_LOW_PRIORITY,
 	SP_NOTIFY_RESP,
+#else
+	SP_RESPONSE = 8,
+	/*
+	 * Note: TISCI documentation indicates "low priority", but in reality
+	 * with a single thread, there is no low or high priority.. This usage
+	 * is more appropriate for TF-A since we can reduce the churn as a
+	 * result.
+	 */
+	SP_HIGH_PRIORITY,
+#endif /* K3_SEC_PROXY_LITE */
 };
 
 /**
diff --git a/plat/ti/k3/include/platform_def.h b/plat/ti/k3/include/platform_def.h
index a56dd3d74..f12fb0b21 100644
--- a/plat/ti/k3/include/platform_def.h
+++ b/plat/ti/k3/include/platform_def.h
@@ -162,12 +162,21 @@
 #define K3_GIC_BASE	0x01800000
 #define K3_GIC_SIZE	0x200000
 
+#if !K3_SEC_PROXY_LITE
 #define SEC_PROXY_DATA_BASE	0x32C00000
 #define SEC_PROXY_DATA_SIZE	0x80000
 #define SEC_PROXY_SCFG_BASE	0x32800000
 #define SEC_PROXY_SCFG_SIZE	0x80000
 #define SEC_PROXY_RT_BASE	0x32400000
 #define SEC_PROXY_RT_SIZE	0x80000
+#else
+#define SEC_PROXY_DATA_BASE	0x4D000000
+#define SEC_PROXY_DATA_SIZE	0x80000
+#define SEC_PROXY_SCFG_BASE	0x4A400000
+#define SEC_PROXY_SCFG_SIZE	0x80000
+#define SEC_PROXY_RT_BASE	0x4A600000
+#define SEC_PROXY_RT_SIZE	0x80000
+#endif /* K3_SEC_PROXY_LITE */
 
 #define SEC_PROXY_TIMEOUT_US		1000000
 #define SEC_PROXY_MAX_MESSAGE_SIZE	56