MSI Claw 8 AI+ vs MSI Claw A8 BZ2EM, Two Amazing Handheld Gaming PCs

This writeup compares two powerful handhelds from MSI, the Claw 8 AI+, and the Claw A8 BZ2EM. Both handhelds vary slightly in price with the AMD-based BZ2EM retailing at $1299 and the Intel-based Claw 8 AI+ retailing for $1199 (USD).

The Claw 8 AI+ comes in several colors, with the original color called “Sandstone” (the color I purchased). It has a somewhat mature but less gamer-looking aesthetic. It also comes in Polar Tempest (white), with both models having a black rear plate. They announced new Glacier Blue and Void Purple colors at CES, and both looked amazing in person. No word on official release date yet for those.

The AMD-based Claw A8 has quite a different look, using angled edges and an aggressive design. This is a newer model, though we’ll discuss performance comparisons later in this review. The launch color is Polar Tempest (white), but MSI teased a handful of more intense colors coming soon. Oddly, neither model comes in pure black.

The Claw 8 AI+ uses an Intel Core Ultra 7-258V (Lunar Lake) CPU, and and Intel Arc 140V (8 Xe2-cores) GPU. The Claw 8 AI+ also includes an NPU with 47 TOPS, which is not too common with handhelds. The only other known handheld with an NPU right now is the ASUS ROG Xbox Ally X, which uses the AMD Ryzen™ AI Z2 Extreme processor.

NPUs aren’t essential for raw gaming performance, but there is potential for Windows to continually leverage benefits of NPUs by offloading some tasks to it. While the MSI Claw A8 BZ2EM also comes with the AMD Ryzen Z2 Extreme (Strix Point), the MSI version does not have the NPU included, hence the non “AI” in the name.

It does of course use the same AMD Radeon™ 890M GPU. Going forward, I will refer to the MSI Claw 8 AI+ (Intel) as the Claw 8, and the MSI Claw A8 BZ2EM as the Claw A8. If you need help remembering, A8 = A for AMD.

Both the Intel Core Ultra 7-258V and the AMD Ryzen™ Z2 Extreme are powerful chips for handheld applications, and in some cases offer very similar performance. This review will focus on raw performance differences in different scenarios, synthetic benchmarks, ergonomics, noise, buttons, etc. Handhelds should not be judged exclusively on raw performance, and this write up should help you decide which of the two is better for you.

SPECS

It’s worth highlighting a couple minor differences. The Claw 8 AI+ comes with 32 GB of DDR5X memory, and at a slightly higher speed of 8,533 MT/s. The Claw A8 comes with 24 GB of DDR5X memory, at 8,000 MT/s. Perhaps this is part of RAM constraints and higher price points, but the older Claw 8 AI+ is certainly the better value when it comes to RAM capacity and bandwidth.

It’s easy to say more RAM is always better, and while I agree, there was no workload or synthetic benchmarking I performed that demonstrated a clear need for more RAM with either of these CPUs.

Storage Test

Although both units have a 1TB NVME drive, the MSI Claw 8 AI+ uses a smaller 2230 size drive. The Claw A8 uses a full size 2280 NVME SSD, which makes potential storage upgrades down the road easier. The 2280 form factor (80mm long) is far more ubiquitous and is usually available at most local computer parts retailers.

Both drives loaded maps and levels seemingly in the same amount of time, so although synthetic benchmarks favor the Intel-based Claw for some tests, and the Claw A8 for others, I wouldn’t worry about the performance delta between the two for gaming. They also have the same 300 TBW endurance rating, so reliability should be comparable as well.

ERGONOMICS AND LAYOUT

I think ergonomics is one of the most important aspects of a handheld. What good are performance and specs if it’s awkward to hold or painful to use for extended periods of time? It’s here that I give a slight edge to the Claw A8. It weighs a little less and the somewhat aggressive angled grips require less grip strength to hold securely.

The Claw 8 feels a little more slippery if I am holding it in a more vertical orientation when laying back/down, where the Claw A8 takes less effort. The design of the Claw 8 feels more robust in hand, making it feel a little more premium at the expense of increased holding fatigue over time.

The small difference shrinks even further if you add a grip case to either of them. I wish Dbrand would make a Killswitch case for either of these handhelds because I love them on others I use. It’s also worth pointing out the somewhat limited aftermarket support for protective cases for either model.

Here you can see a more curved side profile on the Claw 8, and a more of an angular design on the Claw A8.

Below, you can see a more open and robust cooling solution on the Claw 8, but not by much. The middle section is is vented on both Claw models, but not actively cooled. Both the Claw 8 and Claw A8 have active cooling on the outer portions of the heatsinks, clearly shown on the Claw A8.

The average width and height is nearly the same on both handhelds, but the Claw A8 is slightly thicker, purely due to the larger and more aggressive grips.

This pictures do a decent job of showcasing some of the ergonomic differences between both units. The last photo shows the angular cuts in the grip design on the A8 model really well, which I feel gives it more security in hand. One other key detail is the right and left edge angle, and placement of outer (left) stick and ABXY buttons (right).

The ABXY buttons on the AMD edition are lower and further outward. As a result, it felt more natural to transitioning from the right thumb stick to the buttons with my thumb. For fast-paced games, this placement can reduce reaction times if you play more competitively on handhelds.

For the more casual gamer, the difference is less of an issue, especially if you have larger hands. There are other qualities I enjoy on the Intel (Claw 8) variant more, so it’s all about trade offs and what’s most important to you.

In general, the Claw 8 AI+ is more tightly packed horizontally, and more spread out vertically when it comes to control spacing. The Claw A8 may feel more natural to some, especially with shorter thumbs. I’ve used the Claw 8 AI+ extensively over the past few months and have enjoyed it regardless.

This is a somewhat crude attempt at creating an overlay with both handhelds. The positioning isn’t perfect, but for visual people out there, I hope this gives you a better approximation on the layout difference between both.

NOISE AND FAN PROFILE

Yes, this is its own category. I first discovered the importance of this with my first high performance handheld purchase last year, the Lenovo Legion Go S running Steam OS. That is a wonderful handheld, but aside from poor battery life, something else immediately stood out. The fan noise. It was aggressive and loud, making it difficult to play late at night at low volumes.

The next handheld I purchased was the MSI Claw 8 AI+, which was the complete opposite. With the Claw 8 on moderate power profiles, it’s whisper quiet. The newer MSI Claw A8 BZ2EM handheld has decent thermal performance and tolerable noise, but it is louder than the Intel version.

The AMD version has a different heat sink design, which partly contributes to the lower weight of the chassis. Additionally, the the AMD Z2 Extreme is allowed to pull a higher total wattage (TDP) than the Intel chip found on the Claw 8 AI+. At full tilt, the AMD-based Claw A8 fans can hit up to 8,000 rpms, while the Intel-based Claw 8 stays below 4,000 rpms. It is still noticeably more quiet than the Legion Go S, so it’s not unbearable, but the Claw 8 AI+ is the most discrete sounding handheld you can get.

SOUND AND HAPTICS

Both the Claw 8 AI+ and Claw A8 have some of the best sounding speakers I’ve experienced on a handheld.

Ranking sound quality from best to worst are as follows (to me):

1. MSI Handhelds
2. ASUS ROG Xbox Ally X (close second)
3. Lenovo Legion Go 2
4. Lenovo Legion Go S

I would give a slight edge towards the 8 AI+ in sound quality, but for overall gaming performance they are quite similar. They may use the same speakers, so this difference in perceived sound quality could come down the the Claw 8 placing the speakers lower on the front faceplate. I love that both models are front facing speakers either way, as I think that’s the preferred layout.

I cannot say the same about haptics however, where both MSI units tend to struggle when compared to other handhelds. Every other handheld I’ve tried seemed to have stronger haptics systems than this.

Haptics can be fun and immersive, but I typically disable haptics on all controllers when I play. I’ll admit when I tried the wonderful haptics found on the ROG Xbox Ally X from ASUS, or the Legion Go 2 and Nintendo Switch 2, they all showcased a greater level of immersion for some games which can be more fun for some.

This is all a personal preference thing, but it’s worth pointing out.

THUMBSTICKS, TRIGGERS, AND BUTTONS

As someone who primarily plays FPS and racing games, thumbstick performance is incredibly important to me. I typically play with some of the best gaming controllers you can buy, using advanced hall effect or TMR sticks.

Hall effect sticks have a huge variance in performance, mainly in the center “dead zone” area. If they aren’t calibrated well, it can negatively impact super fine aim adjustments, which can be frustrating with skilled players. A huge pro of hall effect sticks though is the lack of eventual stick drift often found in traditional thumbsticks known to use ALPS-based sensor systems.

I love the feel of a good ALPS stick. They have smooth resistance and give really nice feedback with small movement. Unfortunately they have a terrible failure rate with drift issues, often requiring replacement controllers.

When you make an expensive handheld PC, Hall Effect and TMR (Tunneling Magnetoresistance) sticks are the way to go. They use magnets instead of potentiometer-based systems, which rarely fail. You don’t simply through away a +$1000 PC because of stick drift. You either have to repair it or deal with it.

MSI went with Hall Effect thumbsticks on both models of the Claw. Although TMR would be even better, it’s not common on handhelds yet. Both models do however have RGB rings around the thumbsticks, customizable through MSI Center M’s software. I wish you could adjust ABXY RGB brightness independently of the thumbstick, but they too light up when RGB lighting is enabled.

MSI’s sticks are smoother and lighter feeling than most other handhelds, which can make it more challenging for precise aiming in competitive FPS. Lower sensitivity settings helped counter this to an extent. The Claw A8 is a touch more firm and therefore easier for me to aim on, but they are similar.

I found the traditional analog sticks on the ASUS ROG Xbox Ally X to be the best for FPS (no surprise), but there’s always the fear of what happens 6 months from now if the sticks begin to drift. Aftermarket kits exist for the ROG Xbox Ally X, but I don’t think everyone is comfortable with that replacement process. With the smaller 7″ screen on the ROG Xbox Ally X, I faced different challenges with FPS gaming.

The buttons, triggers, and bumpers are all great on both MSI Claw models. The rear buttons on the Claw A8 are more pronounced and tactile feeling, while the Claw 8 is more flat and integrated into the rear faceplate.

Both have a near identical click feel and sound, so it’s more of a design difference than anything else. You can of course customize what each rear button does, whether mapping it to a controller input (ABXY, d-pad, etc.), a Windows input, or disabling them entirely.

The bumpers and triggers feel slightly different. The Claw 8 has a softer and less tactile bumper press, while the Claw A8 has a more clicky, and louder bumper press. If you prefer a very sure tactile feeling of bumper actions, the Claw A8 is the way to go. I prefer the Claw 8 model for late night gaming, as it doesn’t risk disturbing the Mrs. as much.

The triggers are hall effect on both, and have a near identical squeeze resistance to them. The Claw 8 has a wider trigger, which feels nice on my larger hands. The Claw A8 doesn’t have a sharp sensation at all, but it is a more narrow and focused pressure point in the inside of your finger.

They both have the same amount of trigger movement, with the Claw 8 AI+ having a more quiet landing at full squeeze when the trigger bottoms out.

The D-pads look and feel identical at the surface level. They have a good balance of resistance and tactility, and I would rank both as average. The Claw A8 has a more consistent, but firm landing, making it feel slightly more precise, but the Claw 8 is not far behind.

Both models seem to use identical ABXY buttons, and both support RGB illumination. They have a softer membrane feel to them for the switch actuation, and should be very familiar to those having experience with Playstation and Xbox controllers.

Both models lack a track pad, something the Legion models and Steamdeck models have. You can set the control mode to “Desktop mode” (MSI Control Center widget) on either unit though, allowing you to use the left joystick to control a mouse on the screen.

PORTS AND I/O

Both units have a nearly identical port layout. MSI puts all ports and connection options on the top edge of the handheld. The power button doubles as a fingerprint reader for both models.

Both have a Micro SD slot and 3.5mm TRRS combo port for wired headsets, headphones, and IEMs. You also get two USB-C ports, with a minor difference. The Claw 8 uses two official Thunderbolt 4 ports (backwards compatible with USB 4, USB 3.2, 3.1, etc.). The Claw A8 uses two USB-4 ports, also rated for 40 Gbps and claim TB4 compatibility.

Both units have the same theoretical port performance, and in practice that’s what I experienced. This essentially comes down to the Intel chip having a built in Thunderbolt controller due to Intel owning the Thunderbolt trademark and technology. AMD simply makes their ports compatible with it.

SCREEN

The screen is wonderful, and identical, on both models. It is an 8″ IPS LCD display, with a 1920×1200 native resolution (16:10 ratio), and a max refresh rate of 120 Hz. Peak brightness arrives at 500 nits, and I find it serviceable most lighting conditions aside from extended use in direct sunlight.

These panels support VRR as well, or variable refresh rate. This highly desired feature with handhelds will adjust the screen’s refresh rate in real time to match the frames per second coming from the game you play, making motion feel smooth even if frames drop from time to time.

VRR must be enabled on screen settings AND in game in order to benefit from this. This screen’s VRR range is 48 to 120 Hz, so your game graphics quality settings must be able to maintain a minimum of 48 FPS to benefit from this. There are third-party tools out there (such as CRU utility) that can extend VRR range down to 30 Hz as well.

One important note – If you enable the on screen display for real time frame rate monitoring on the Claw 8 AI+, VRR stops working correctly. I suggest leaving OSD off if you are playing with VRR enabled.

I think the 8″ screen size is close to perfect for a handheld device. It may not sound like much, but going from a 7″ to an 8″ screen increases your total screen volume by about 30%! It’s wonderfully immersive, and makes it easier to read the UI of certain games as well. While larger screens are always welcome, it can come at the sacrifice of weight, comfort, travel portability, and battery life – all important things to consider for a handheld.

The displays have a glossy finish, which is reflective in many environments. This is common, especially considering they are both touch screen displays (a must for a handheld). I find the Lenovo 8″ LCD screen to have a touch more color saturation with default settings, but overall I think the 8″ screen found on the Claw models is excellent and I have no real complaint.

BATTERY LIFE AND TDP SETTINGS

Both MSI Claw 8 AI+ and Claw A8 models use the same 80 WH battery. This ties them with the highest capacity available on mainstream handhelds. They occasionally trade off efficiency advantages depending on the game you are playing.

For example, if you target 45 FPS on Call of Duty Black Ops 7, the Claw A8 BZ2EM edition with the AMD chip will last longer. If you target 60 FPS in ARC Raiders, the Claw 8 AI+ with the Intel chip will last longer.

The AMD version will typically always run out of battery first based on my experience. The Intel model is clamped at 30W max TDP sustained, while the AMD variant allows you to run at a sustained 35W TDP spec, providing extra performance at the cost of noise and battery life. On more moderate settings I find the Intel version to have a small edge but it’s not significant.

MSI Center M’s quick settings allows you to set manual TDP configs for either model, and they also provide you with two pre-defined scenarios called AI Engine, and Endurance. MSI chose different power levels for both variances in either mode. After testing a handful of games with all available power settings, I observed the maximum sustained TDP draw for each chip/Claw model.

If you had both handhelds side by side and compared the MSI Claw 8 AI+ to the Claw A8 BZ2EM, using the default power profiles typically resulted with the Z2 Extreme chip outperforming the Intel chip in several games.

When we manually configure the TDP to 20W on both models, performance is extremely similar, often varying due to the game being tested and if Image Scaling was used (AMD FSR or Intel XeSS). Running either of these default power modes maintains reasonable fan noise, with the Claw 8 AI+ being more quiet in all scenarios.

The Endurance mode on the Claw 8 provides greater battery life and less noise at the expense of performance when compared to the Endurance mode on the Claw A8. It also enables Intel’s “Endurance” profile on their graphics settings which caps all games FPS to just 30.

Yes, even on a game like Hades II, the FPS cap further extends battery life by asking the GPU to render less frames. Although a 12W TDP could generate more frames, this puts a cap on it. Some may think 30 frames per second is simply byproduct of having a very low TDP setting, but that’s not always the case.

You can turn this frame limit setting off manually in the Intel graphics settings by changing the endurance mode to 40, 60 FPS, or even unlimited. I found that plugging in the Claw 8 AI+ to dedicated power ended up resetting/breaking this setting, so I’d have to manually change the Endurance cap each time to “re-unlock it” for more performance. Thankfully, a tool exists to fave me some time…

TOOTH N CLAW APP FOR CLAW 8 AI+

There is an app for the MSI Claw 8 AI+, now available on the Microsoft App Store, called “Tooth N Claw”. This is a must-install feature for the Intel model and is by far one of the most handy utilities I’ve ever tried on a handheld.

It integrates perfectly with the Microsoft Full Screen Experience (FSE) as well. Once installed, pressing the shortcut button on the right side of the screen launches the FSE quick settings, including the widget dashboard of MSI Center M (used for TDP, OSD, control mode, etc).

Press the right bumper several times to navigate to two new sub menus for Tooth N Claw. The first menu is called Color Remaster. I bumped up Saturation to 60, and left it at that. You can also change brightness, contrast, and image sharpening settings here as well.

Pressing the right bumper once more takes you to the Performance menu. This is where the magic happens. You can customize the CPU Boost Mode, which changes how the Intel chip handles a steady state vs peak.

If you set boost mode to “Disable”, it disables CPU boost, allowing more thermal headroom and power for the GPU. This may make emulators or Windows-focused apps feel slower, but it can in some cases make GPU performance better. Two other settings I used often was the Display Resolution drop down, which makes it far easier to change screen resolutions without navigating to Windows settings.

If you recall my issue with endurance mode in the TDP settings above, the FPS cap can be a bother to fix when changing power profiles and charging the Claw 8. Thankfully, Tooth N Claw allows you to quickly toggle the Target Framerate from 30 to 40, 60, and Unlimited without needing to exit your game or launcher. For some games, I loved using the 60 fps mode with the low 12W TDP setting. The Claw 8 AI+ is essentially silent, and I can easily play for 4-6 hours on a single charge.

GAME PERFORMANCE & BENCHMARKS

I took the time to test a handful of games with different configurations. I tried different TDP levels, and different graphics settings to see how each handheld handles each particular game. Some tests are from games I had a personal interest in testing, along with some new games I was excited to play on handheld.

Test Process

Some of these results are from actual in game tests, and not traditional benchmarks. For those tests, I loaded the same spot several times and averaged three readings to maintain consistency. For official benchmarks, these numbers represent an average of 3 benchmarks per handheld in each presented result.

The handhelds were fully charged, then disconnected from dedicated power, run at the highest available TDP setting, and also at a 17W manual TDP setting. I did not observe any performance impact on multiple handhelds whether plugged in or not, but I wanted to keep the test process the same for all benchmarks just in case.

There are a few graphs that have “XeSS/FSR” in part of the title. These are very unique benchmark presentations, in which the MSI Claw 8 AI+ was using XeSS – Balanced profile, while all the AMD-based handhelds were using the FSR 3 Profile. Intel GPUs benefit more from XeSS image scaling than AMD.

As a result, the gaps/differences can vary. I prefer the image quality more with XeSS image scaling, but either option can increase frame rates with all other settings being equal. I also included the same benchmarks where all handhelds were running XeSS, and where all handhelds were running FSR – both in balanced mode. I also ran the same tests with no image scaling enabled at all, again to be as consistent and in depth as possible. It’s interesting to see how each configuration impacts each handheld with various games.

Other Handhelds Tested

To help clear any confusion, these are the specs of other models tested for the benchmarks below:

• ASUS ROG Xbox Ally X – AMD Ryzen AI Z2 Extreme (50 TOPS NPU), 24GB RAM, 1TB SSD
• Legion GO 2 – AMD Ryzen Z2 Extreme, 32GB RAM, 1TB SSD
• Legion Go S – AMD Ryzen Z1 Extreme, 32GB RAM, 1TB SSD, Steam OS

ARC Raiders

First up is ARC Raiders, a game I personally love. ARC Raiders does not have a benchmark tool, and playing the game drops into live events with weather and different environment mods. To produce a consistent test, I loaded the Practice Range, and recorded the frame rates over time in the exact same spot on all handhelds. The Practice Range does not load other players, and there are no other map/world modifiers, so it’s always the same. This produced the most consistent result, typically within 1-3 FPS per test.

The ASUS ROG Xbox Ally X had a stronger 1080p performance at max TDP. ASUS allows an even higher max TDP for their Z2 extreme than other models using the same chip, so this is more of a byproduct of raw peak power than it is other factors.

If you look at the 17W TDP limit results, the edge shifts back to the MSI models with one anomaly for the Claw A8 at 17W with XeSS. At higher resolutions, the Z2 Extreme is again a solid performer in raw raster, while the Claw 8 AI+ benefits significantly from XeSS image scaling.

As mentioned before, XeSS looks better (typically) than FSR, so I give the overall real-world performance edge to the MSI Claw 8 AI+. Nevertheless, all handhelds provided an enjoyable gaming experience without needing to max out power levels. If you leave the frame counter off, you’d be hard pressed to spot any difference amongst these models.

Call of Duty: Black Ops 7

Next up we have Call of Duty Black Ops 7, which is an AMD-optimized game. In essentially all settings, the MSI Claw A8 outperforms the MSI Claw 8 AI+. There is enough optimization that the Claw A8 in 17W TDP mode is almost as fast as the Claw 8 AI+ at max TDP.

Other AMD-based handhelds also performed well in this test, with the omission of the Legion Go S due to running Steam OS. Some games with certain anti-cheat tools do not work with Steam OS, Call of Duty and Battlefield included. As much as I enjoy the simplicity of Steam OS, recent updates to Windows 11 and their Full Screen Experience (FSE) allows me to at least attempt to run every game I purchase.

Also noteworthy on these results is the worse 1% and 5% lows on the Intel-based Claw 8 compared to the AMD handhelds. These dips to low frame rates, especially when below the VRR frequency range of the display, can lead to occasional frame stutters and break immersion. As a competitive player, I prefer to avoid this whenever possible. If your main game is Call of Duty, it’s a clear edge towards the Z2 Extreme for performance and overall enjoyment.

Cyberpunk 2077

Cyberpunk 2077 is a more neutral benchmark, not showing a clear preference on GPU. All handhelds performed comparably to each other, with even the older Z1 Extreme found on the less expensive Legion Go S performing well. Part of the performance parity may be due to the efficiency of Steam OS working particularly well with this game, the Linux-based AMD driver, or other factors. At the end of the day, the results were great.

One noteworthy result was the MSI Claw 8 AI+, using the Intel GPU, had some clear performance advantages with the lower 17W TDP setting. This was amplified further with XeSS image scaling enabled, set to balanced mode. For some odd reason, after multiple tests, the Claw A8 seemed to have the lowest 17W performance in this benchmark.

It may be caused by more aggressive power limiting in 17W TDP mode. Setting the Claw A8 FPPT to 20W instead of the hard cap at 17W across the board produced better results, but I did not have to do this with other Z2 Extreme models. The variance could be from BIOS power management or even the latest driver available by MSI compared to Lenovo or ASUS.

At 1200P with max TDP settings, the MSI Claw handhelds both top the charts without image scaling. FSR performs better on AMD-chips, and the Intel XeSS image scaler further boosts the performance on the Claw 8 AI+, showing similar performance uplifts observed before.

Hades II

Moving on to more handheld-friendly games, Hades II was a surprise hit for me. The game is absolutely stunning to play, super light on resources, and it feels like it was meant for handheld gaming. You can expect great performance results regardless of which handheld you buy, including less powerful and more affordable models. The edge typically favored all AMD-based handhelds, but I’m not complaining about 300 fps either way.

At 1200P, the raw power of the Z2 Extreme shows, maintaining almost the maximum FPS the game engine allows at a more crisp resolution. The ASUS ROG Xbox Ally X is omitted from the 1200P results due to having a smaller 1080p screen and the game not allowing a forced resolution beyond the display’s spec. It’s safe to assume comparable performance considering how tight these results were.

LEGO Star Wars : The Skywalker Saga

LEGO Star Wars : The Skywalker Saga, was a challenging title to benchmark… at first. I learned that having a touch-screen enabled Windows device running this game had poor performance. This was only discovered when I eventually tested the Legion Go S running Steam OS, which produced significantly higher frames per second.

After doing some research, a Reddit post led me to a custom file that patched the game, along with other games built on this engine. It’s called TTTouchFix, by Nezarn (https://github.com/Nezarn/TTTouchFix/releases). Pasting this dll in either the x86 or x64 game folder (depending on game), immediately fixed my low frame rate issues. Just make sure to copy this over while the game is closed.

Now that it was fixed, it was time to spawn back in and test. I spawned in the exact location for each handheld with every setting, looked at the same point, then captured the FPS over a 30 second average. This process was repeated several times to ensure the most consistent and accurate results.

At the end of the day, all handhelds performed well, and this game was also a blast to play on a handheld. The MSI Claw 8 AI+, using the Intel Arc 140V GPU, topped 3 of 4 benchmarks. The Claw A8, using the Radeon 890M, had nearly identical performance, and better performance at 1080P 17W TDP mode.

In general I think the Intel-based systems work better closer to 19-20W TDP for similar results due to how the CPU is packaged, due to having an onboard memory controller. The Intel chip consumes less total system power at the 17W configuration, which can lead to slightly less performance in some games, but ultimately a better battery life in these modes.

Hogwarts Legacy

Hogwarts Legacy is an enjoyable game on handheld as well, and it’s very easy to customize the graphics to get the most out of whatever device you own. I ran all benchmarks with the medium graphics preset. The only changes I made were 1080P vs 1200P, the TDP mode of full power vs 17W, and the image scaling options. Image Scaling was either disabled, on FSR 3 Balanced, or XeSS Balanced. I did not use frame generation for these benchmarks, or any prior benchmarks.

Following the same testing methodology as LEGO Star Wars, I spawned to the same location and focused the view on the same point. The frame rates occasionally deviated by a few frames at a time, but nothing drastic. This game required several launches of the game in multiple settings to get consistent results.

Regardless of how many times I reset the game settings, restarted the handheld, etc., for some reason the Lenovo Legion Go 2 had poor FSR performance in 1080P. It may came down to a combination of the BIOS, drivers, and TDP constraints, or even more aggressive thermal throttling than the other models here. Oddly, it performed better at 1200P with FSR enabled. In general, AMD chips performed slightly better on this game, but the Intel based Claw 8 AI+ had solid performance with XeSS enabled.

XeSS is a great image scaler for games that support it, as I find it looks better than FSR with similar scaling settings when it comes to detail preservation, graininess, and texture shimmer. So even though AMD led most results here, I think the Intel-based handheld had the most enjoyable performance at reasonable (quiet) TDP settings. For a game like this, and even ARC Raiders, I would try changing your scaler options from FSR to XeSS to see which one you prefer, even if you have an AMD GPU.

USING AN EXTERNAL GPU

To see how far we could push both MSI handhelds with performance from their CPU, I paired them to an external GPU. This was made possible by Razer, who were kind enough to send over their new Core X V2 E-GPU ($349.99 without GPU or PSU). They also sent an XPG Core Reactor 2 850W power supply to go with it, which is an option on their site when purchasing.

This E-GPU chassis is a beast, and handled my RTX 5090 with a little extra room for future card growth. It also came with a side fan to help move air out of the chassis, though the entire lid is passively vented. Your GPU fans alone should be enough to expel enough air for safe temps, but I like the quiet fan inclusion either way. In typical Razer design, you have the customizable RGB Razer logo up front. You do not need to install Synapse to use this chassis though, making it full plug and play.

I decided to test external graphics with the MSI RTX 5090 Gaming Trio OC – an absolute unit of a GPU. I then used the included Thunderbolt cable, connected it to each MSI handheld, and starting running some tests.

The Claw 8 AI+ has true Thunderbolt 4 ports (two of them), and I expected the connection to be seamless. For some reason, the only way my external test monitor (Samsung G80SD 32″ 4K 240Hz QD-OLED) would work was to boot the handheld and log in, then unplug and immediately plug the Thunderbolt cable back in. Once this handshake happened, it was smooth sailing. I tried installing the official Intel Thunderbolt software, but it did not change the behavior.

The MSI Claw A8 immediately recognized and enabled the monitor without any additional software needed, a very pleasant surprise.

These tests were quite different from each other. For Arc Raiders, the first 1080P test was using the same Medium graphics preset used for the handheld benchmarks before. Both handhelds pulled a respectable 68 FPS on that mode, wire free.

Trying the same settings with the E-GPU connected, both handhelds more than doubled their performance. While high frame rates are expected from an RTX 5090, max frame rates can be lower due to the CPU used (to an extent). A much older desktop processor will have lower frames than a modern AMD Ryzen 9800x3D for example, even when using something like the 5090 in 4K.

Pushing the graphics limits further, I went to an extreme test. With this being an RTX 5090, I chose DLSS for my image scaler using the DLSS Transformer model. Ray Tracing was set to Dynamic – Epic, the highest setting. The graphics preset was set to Cinematic, also the highest preset.

Normally, this would hammer the GPU, and that would become the bottleneck. I expected frame rates to be comparable to each other due to that, but the AMD Ryzen Z2 Extreme on the Claw A8 had a surprising uplift. The Z2 Extreme uses a unique core architecture of 3 big and 5 compact cores (using different process nodes). AMDs total thread count is 16 vs Intel’s total thread count of 8, but it’s hard to say how much is impacting the difference here beyond other potential factors.

Cyberpunk 2077 tests resulted in comparable performances between both MSI handhelds. In general, the Claw 8 AI+ had a tiny edge in performance. The highest fidelity setting resulted in the Claw A8 pulling ahead again with a smaller difference than the ARC Raiders test. It’s quite easy to find a good middle ground for graphics settings that provides wonderful fidelity with a high frame rate, as seen in a couple of these tests.

FRAME GENERATION

One additional boost to performance you can benefit from is something called frame generation. Frame generation, or “frame gen” does exactly what the name implies. It generates extra frames, artificially, between frames to improve smoothness with gameplay.

A 2x frame gen can make your 30 fps game look and feel like a 60 fps game. It’s not entirely as simple as this, because poor baseline frame rates may still not be enough to result in a smooth gaming experience, even with frame gen enabled. You may also feel higher control latency, or some form of delay from what your intended user inputs are versus how the game responds. Many competitive players avoid frame gen for this reason.

There is also the potential for visual artifacts or odd rendering issues in game. Fine details may look weird, or have an odd jelly effect with fast camera movement. You may also have some shimmering issues or blurriness in some scenes as well.

So why use it? Well, a lot of these potential downsides are just that… potential. In some cases, frame gen works wonderfully, and can greatly improve your gaming experience. Nvidia, AMD, and Intel all have different frame generation models. With these handhelds being AMD and Intel, those are your two models to pick from, though newer XeSS (2.1 and up) frame generation models can still work on AMD.

Confused yet? Well, this is why I won’t have dedicated benchmarks for this. Frame Generation is incredible, and Intel seems to have the upper hand with frame generation with handhelds right now. The latest update from Intel’s driver suite (XeSS 3.0) allows up to 4x frame gen! It sounds wild, but somehow it works. I was able to use XeSS frame gen on both Cyberpunk 2077 and Hogwarts Legacy with great results. With the right settings, I had no noticeable visual drawbacks. What I did get was a massive performance boost, even with lower TDP settings.

I pushed this even further by using moderate frame generation on Battlefield 6. Pushing 80+ FPS on the medium preset with XeSS enabled, and a 25W TDP cap on the Claw 8 AI+ is impressive. I had less perceivable input lag on XeSS than AMD’s frame generation, leaving me to disable it on AMD-based models in Battlefield 6 if I actually wanted tolerable aim response.

DRIVERS & SUPPORT

In many cases, these handhelds can be plug and play. After setting up your user account, it’s typical to download Windows and app store updates. I did that for both devices, and used MSI’s Center M software to check for updates. To my surprise, AMD Driver updates for the GPU are few and far between.

This isn’t exclusive to MSI either. Lenovo and ASUS both had older AMD drivers as well. When AMD releases new GPU drivers, the approval process for handhelds goes through the manufacture for validation and compatibility testing.

AMD’s driver support process for handheld chips isn’t great, and seemingly got worse over the past year. So while their older drivers have excellent compatibility with new and old games, updates can take many months before users get them. Some owners of older handhelds reportedly stopped receiving updates.

This is a unique positive for the MSI Claw 8 AI+. Although Intel’s ARC GPUs don’t always play nice with some games, particularly older ones, the Claw 8 allows you to install Intel Driver Assistant. This amazing program handles chipset, WiFi, Bluetooth, and those important GPU driver updates.

I’ve installed several driver updates since purchasing the Claw 8 AI+ last year, and performance has improved each time. The recent Frame Generation update mentioned above created a whole new performance level the 8 AI+ could not achieve before. This bodes well for long term ownership of the Claw 8 AI+ unless something changes from Intel or AMD.

In almost all situations, I had no issues with the older AMD drivers. I had a crashing bug on Cyberpunk 2077 that affected all other AMD based handhelds except the Claw A8 for some reason. On my ASUS ROG Xbox Ally X and Lenovo Legion Go 2, the game simply won’t load anymore without crashing at launch.

This affected both handhelds after the most recent Windows or Cyberpunk 2077 update. All three AMD Z2 Extreme handhelds have different GPU drivers at the time of writing this, yet they are all on the latest one supported by the manufacturer. This is a small, but unfortunate issue to have to consider for long term ownership of your handheld gaming device.

FINAL THOUGHTS AND EXPERIENCES

To simplify this as much as possible, both handhelds perform extremely close to each other. A few games favor AMD, and a few favor Intel. I appreciate the raw performance of the Z2 Extreme chip found on the Claw A8, but there is a small caveat of slightly shorter battery life, and more fan noise. On games like Call of Duty, the 1% and 5% lows are much better on the Claw A8, and the game was noticeably smoother and enjoyable to play.

I think MSI made some smart choices with the shape of the new Claw A8 as well. Though I like the sleeker look of the Claw 8 AI+, the Claw A8 was a little more enjoyable to play in vertical orientations thanks to the shape of the handles. The A8’s thumb sticks seem slightly more stiff as well, lending itself to more consistent aiming in first person shooters. You can see that MSI took their experience from the 8 AI+ and older models to further refine the overall design and performance of the Claw A8.

Although the Claw 8 AI+ is slightly more slippery to hold, a case can help a lot. What I most enjoy on the Claw 8 AI+ is the efficiency. It has a robust cooling solution and overall pulls less power, even at full load. The end result is it consistently being the most quiet handheld I’ve ever used, which is amazing for late night gaming. Even the buttons are quiet!

With wonderful driver support and shockingly good XeSS Frame Generation performance, the Claw 8 AI+ has only improved in the months I’ve had it. Sometimes I wish it was a little more refined, mainly with the thumb sticks and haptics system, but I’m happy either way.

Both handhelds had rock solid performance everywhere else. The WiFi performance was excellent, and I never had a dropped wifi connection on either handheld. The MSI Center M isn’t the prettiest software, but it worked well for me. I only had a couple times where the Claw A8 “Gamepad mode” wasn’t detected by the game. The joystick and ABXY buttons inputs weren’t registering occasionally on a few games (mainly ARC Raiders). Toggling from desktop mode back to gamepad mode fixed this issue though.

I will be releasing a video comparing these two MSI units to other models soon. MSI will fair well in that comparison due to several reasons. You get a near perfect sized 8″ screen, which I know isn’t OLED but I appreciate the lower comparable price point. MSI provides Intel and AMD options, great build quality, class leading speakers, great battery life, and more. With Lenovo increasing their Legion Go 2 price (dramatically and painfully), the MSI handhelds are actually in a competitive price range at the time of writing this.

After testing 6 handhelds for the past few months, I can confidently say the perfect handheld doesn’t exist. I can also say that MSI has come the closest so far.