diff --git a/Docs/source/design.rst b/Docs/source/design.rst
index 018c61f26..cac914865 100644
--- a/Docs/source/design.rst
+++ b/Docs/source/design.rst
@@ -41,7 +41,13 @@ and the generic solvers:
 
 * ``util/``: linear algebra solvers and other routines.
 
-Design Philosophy
+
+.. note::
+
+   All quantities are assumed to be in CGS units, unless otherwise
+   specified.
+
+Design philosophy
 =================
 
 Any application that uses Microphysics will at minimum need to
@@ -67,7 +73,30 @@ solvers (ODE integration for the network and Newton-Raphson root
 finding for the EOS) are separated from the specific implementations of
 the microphysics.
 
-.. note::
 
-   All quantities are assumed to be in CGS units, unless otherwise
-   specified.
+
+GPU considerations
+==================
+
+.. index:: GPUs
+
+All of the Microphysics routines are written to run on GPUs.  This is
+enabled in application codes by using the AMReX lambda-capturing
+mechanism (see the [AMReX GPU
+documentation](https://amrex-codes.github.io/amrex/docs_html/GPU.html)
+for more information).
+
+This means leveraging the AMReX data-structures, macros, and
+functions.  The unit tests (see :ref:`sec:unit_tests`) provide a good
+reference for how to interface the Microphysics solvers and physics
+terms with an AMReX-based code.
+
+There are a few places where Microphysics behaves slightly differently
+when running on a CPU vs. a GPU:
+
+* In the VODE integrator, we disable Jacobian-caching to save memory.
+  See :ref:`ch:networks:integrators`.
+
+* In general we disable printing from GPU kernels, due to register
+  pressure.  Some output can be enabled by compiling with
+  ``USE_GPU_PRINTF=TRUE``.
diff --git a/Docs/source/unit_tests.rst b/Docs/source/unit_tests.rst
index 6c01f664c..eb8589313 100644
--- a/Docs/source/unit_tests.rst
+++ b/Docs/source/unit_tests.rst
@@ -1,3 +1,5 @@
+.. _sec:unit_tests:
+
 **********************
 Overview of Unit Tests
 **********************