Every gaming handheld on the market right now runs on a chip that was designed for something else. The Steam Deck uses a custom AMD APU originally architected for laptops. The ROG Ally runs Ryzen Z1, which is functionally a Phoenix laptop die in a smaller package. Even the MSI Claw — Intel’s own first attempt — shoved a Meteor Lake laptop chip into a handheld form factor and hoped for the best. It didn’t work. The thermals were bad, the battery life was worse, and reviewers buried it. Intel just admitted that approach was a mistake — and at Computex 2026, it unveiled the fix: Arc G3 and Arc G3 Extreme, the first processors in the industry purpose-built from the ground up for gaming handhelds.
This Isn’t Recycled Laptop Silicon — It’s a Dedicated Architecture
The distinction matters more than Intel’s marketing will tell you. When you take a laptop chip and cram it into a handheld, you inherit the laptop’s thermal envelope, its power management quirks, and its memory controller design — all optimized for a 15-inch chassis with a fan the size of your palm. Handhelds don’t have that luxury. They run in a 15W-30W window, they need to sustain performance for hours on a 50Wh battery, and they dump heat through a chassis you’re holding in both hands.
Intel’s Arc G-Series is built on Panther Lake — the same architecture behind Core Ultra Series 3 — but specifically tuned for handheld power and thermal constraints. Both the Arc G3 and Arc G3 Extreme share a 14-core CPU layout: two performance cores for heavy single-threaded loads, eight efficiency cores for sustained gaming, and four low-power cores for background tasks and idle power savings. The difference between the two chips is the GPU.
Arc B370 vs. Arc B390: The GPU Split That Actually Matters
The standard Arc G3 ships with 10 Xe3 GPU cores clocked at 2.2 GHz, carrying the Arc B370 label. The Arc G3 Extreme steps up to 12 Xe3 cores at 2.3 GHz under the Arc B390 designation. Both support up to 24GB of LPDDR5x RAM and hardware-accelerated ray tracing via Xe3’s native RT units.
To put that in context: AMD’s Ryzen Z2 Extreme — the chip inside the upcoming ROG Ally 2 — uses an RDNA 3.5 GPU with 16 compute units. Intel’s Arc B390 has 12 Xe3 cores, but direct CU-to-core comparisons are misleading. What matters is real-world frame rates, and Intel is claiming parity or better performance per watt against Ryzen Z2 in the 15-25W range that handhelds actually operate in. Independent benchmarks will settle this, but Intel is confident enough to put its name on dedicated silicon — which means internal testing was strong enough to justify a chip that has no fallback market if it flops.
The Partners Tell You Intel Isn’t Bluffing
The real signal isn’t the spec sheet — it’s who’s building devices around it. Acer’s Predator Atlas 8 is already confirmed with an 8-inch FHD+ 120Hz display, up to 24GB of LPDDR5x, and 1TB NVMe storage. MSI is back with the Claw 8 EX AI+, which is a tacit admission that the original Claw was a misstep. OneXPlayer is onboard. And Intel says more OEM announcements are coming through the summer.
This is significant because the first Claw scared off partners. OEMs don’t commit engineering resources to a chip platform unless they’ve seen sustained benchmark runs that justify the investment. The fact that Acer and MSI are both shipping devices in June 2026 — not “later this year,” not “holiday season” — tells you the silicon has been validated for months.
Follow the Money: Why Intel Built a Chip for a Market That Barely Exists
The global gaming handheld market shipped roughly 8 million units in 2025, according to IDC estimates. That’s tiny compared to the 200+ million laptops sold annually. So why would Intel design, validate, and fab a chip for a market one-twenty-fifth the size of its bread-and-butter business?
Because the growth curve is vertical. Steam Deck moved 5 million units in its first two years. The ROG Ally sold over 1 million in six months. Lenovo Legion Go, MSI Claw, Ayaneo — the category is multiplying. Analysts project the handheld PC market hits $12 billion by 2028, and whoever owns the silicon standard in this segment gets to define the platform.
AMD currently owns roughly 70% of the handheld chip market through Ryzen Z1 and Z2. That’s the kind of dominance Intel used to enjoy in laptops before Apple, Qualcomm, and now Nvidia started eating its lunch. Intel can’t afford to lose another form factor. The Arc G3 isn’t just a product — it’s a territorial claim.
Who Gets Hurt: AMD’s Monopoly, Valve’s Strategy, and Your Wallet
AMD is the obvious loser if Arc G3 delivers. AMD has been the default handheld chip supplier because there was no credible alternative — not because Ryzen Z2 is unbeatable. If Intel achieves comparable gaming performance at lower power draw, OEMs will split their lineups across both suppliers, and AMD’s pricing leverage evaporates overnight.
Valve faces a subtler problem. The Steam Deck 2 is widely expected to run a custom AMD chip. If Intel handhelds start outselling third-party AMD devices, Valve’s hardware moat shrinks — because the best handheld gaming experience might no longer be exclusively tied to Steam’s ecosystem. Intel handhelds run full Windows 11, which means Game Pass, Epic, Battle.net, and Steam all coexist. Valve’s walled garden becomes one option among many.
For consumers, this is unambiguously good. Competition means lower prices, better battery life, and faster innovation cycles. The original MSI Claw launched at $799 with mediocre performance. The new Claw 8 EX AI+ is expected in a similar price range but with dramatically better efficiency. That’s what happens when a chip is designed for your device instead of borrowed from someone else’s.
The Verdict: Intel Just Made Its Most Honest Product Decision in Years
Intel could have shoved another laptop chip into a handheld chassis and called it a generation update. That’s what everyone expected after the Claw disaster. Instead, it did the expensive thing: it designed purpose-built silicon, convinced OEMs to trust it again, and shipped devices within months of the announcement.
The Arc G3 Extreme won’t outsell Nvidia’s RTX Spark or Apple’s M5. It won’t save Intel’s stock price. But it might be the most strategically honest product Intel has released in years — an admission that recycling laptop silicon for every form factor doesn’t work, and that the gaming handheld market is worth fighting for on its own terms. Whether the performance backs up the promise depends on real-world testing. But the decision to build a dedicated chip? That’s already the right call.
