diff --git a/Core/NativeClient/WebGPU/GPU.lua b/Core/NativeClient/WebGPU/GPU.lua
index 7265329e..49e11a5f 100644
--- a/Core/NativeClient/WebGPU/GPU.lua
+++ b/Core/NativeClient/WebGPU/GPU.lua
@@ -11,6 +11,7 @@ local GPU = {
 	MAX_VERTEX_COUNT = 200000, -- Should be configurable (later)
 	MAX_TEXTURE_ARRAY_SIZE = 32,
 	MAX_BUFFER_SIZE = 256 * 1024 * 1024,
+	MAX_UNIFORM_BUFFER_BINDING_SIZE = 65536,
 }
 
 function GPU:CreateInstance()
@@ -78,18 +79,15 @@ function GPU:RequestLogicalDevice(adapter, options)
 				maxUniformBuffersPerShaderStage = 1, -- Camera properties (increase for material, soon?)
 				maxSampledTexturesPerShaderStage = GPU.MAX_TEXTURE_ARRAY_SIZE,
 				maxSamplersPerShaderStage = GPU.MAX_TEXTURE_ARRAY_SIZE,
-				maxUniformBufferBindingSize = 65536, -- DEFAULT
+				maxUniformBufferBindingSize = GPU.MAX_UNIFORM_BUFFER_BINDING_SIZE,
 				maxBindingsPerBindGroup = 2, -- Max. allowed binding index
 				maxDynamicUniformBuffersPerPipelineLayout = 1,
 				minStorageBufferOffsetAlignment = 32,
-				minUniformBufferOffsetAlignment = 32,
+				minUniformBufferOffsetAlignment = ffi.sizeof("mesh_uniform_t"),
 			},
 		}),
 	})
 
-	assert(supportedLimits.limits.minUniformBufferOffsetAlignment <= 32, "Dynamic uniform headaches will ensue")
-	self.minUniformBufferOffsetAlignment = supportedLimits.limits.minUniformBufferOffsetAlignment
-
 	local requestedDevice
 	local function onDeviceRequested(status, device, message, userdata)
 		local success = status == ffi.C.WGPURequestDeviceStatus_Success
@@ -129,11 +127,4 @@ function GPU:RequestLogicalDevice(adapter, options)
 	return requestedDevice, deviceDescriptor
 end
 
-function GPU:GetAlignedDynamicUniformBufferStride(uniformStructSizeInBytes)
-	local step = self.minUniformBufferOffsetAlignment
-	-- More headaches if the dynamic uniforms (e.g., widget transforms) are smaller than the minimum stride...
-	local divide_and_ceil = uniformStructSizeInBytes / step + (uniformStructSizeInBytes % step == 0 and 0 or 1)
-	return step * divide_and_ceil
-end
-
 return GPU
diff --git a/Core/NativeClient/WebGPU/Pipelines/WidgetDrawingPipeline.lua b/Core/NativeClient/WebGPU/Pipelines/WidgetDrawingPipeline.lua
index e5654bb7..58576799 100644
--- a/Core/NativeClient/WebGPU/Pipelines/WidgetDrawingPipeline.lua
+++ b/Core/NativeClient/WebGPU/Pipelines/WidgetDrawingPipeline.lua
@@ -4,6 +4,7 @@ local webgpu = require("webgpu")
 
 local Device = require("Core.NativeClient.WebGPU.Device")
 
+local GPU = require("Core.NativeClient.WebGPU.GPU")
 local UniformBuffer = require("Core.NativeClient.WebGPU.UniformBuffer")
 
 local binary_not = bit.bnot
@@ -12,7 +13,7 @@ local sizeof = ffi.sizeof
 
 local WidgetDrawingPipeline = {
 	WGSL_SHADER_SOURCE_LOCATION = "Core/NativeClient/WebGPU/Shaders/UserInterfaceShader.wgsl",
-	MAX_WIDGET_COUNT = 2048, -- The default maxUniformBufferBindingSize allows for this many without optimizing further/removing padding/using other buffer types or even hardware instacing for the UI
+	MAX_WIDGET_COUNT = GPU.MAX_UNIFORM_BUFFER_BINDING_SIZE / sizeof("mesh_uniform_t"),
 }
 
 function WidgetDrawingPipeline:Construct(wgpuDeviceHandle, textureFormatID)
diff --git a/Core/NativeClient/WebGPU/UniformBuffer.lua b/Core/NativeClient/WebGPU/UniformBuffer.lua
index 5b168b4a..82d1cafe 100644
--- a/Core/NativeClient/WebGPU/UniformBuffer.lua
+++ b/Core/NativeClient/WebGPU/UniformBuffer.lua
@@ -66,8 +66,9 @@ local UniformBuffer = {
 		} water_uniform_t;
 		typedef struct PerMeshData {
 			float translation[2]; // 8
-			float padding[6]; // 32
-			// Total size must be at least minUniformBufferOffsetAlignment bytes large (with 16 byte alignment)
+			float padding[62]; // 256
+			// Struct size must be aligned with minUniformBufferOffsetAlignment (GPU limit)
+			// Padding needs to be updated whenever the struct changes!
 		} mesh_uniform_t;
 	]],
 }