3 Minutes
Samsung is preparing a major shift in its mobile silicon strategy: after years of integrating AMD GPUs into Exynos chips, the company is reportedly developing a fully custom graphics architecture to power future devices, starting with the Exynos 2800 and the Galaxy S28 lineup.
Why Samsung is building its own GPU
Samsung’s collaboration with AMD began in 2019 and produced the Exynos 2200 in 2022, followed by 2400 and 2500 models and even AMD-based graphics in midrange chips like the 1480 and 1580. Those GPUS brought stronger mobile graphics performance and features such as ray tracing to Samsung phones.
Still, multiple reports suggest Samsung believes relying on a third-party GPU limits how deeply it can integrate advanced on-device AI and achieve flawless software optimization. A custom GPU gives Samsung control over the entire stack — hardware, firmware, and drivers — letting the company tune performance, power use, and AI acceleration for its specific goals.
Think of it like what Apple did with custom graphics for iPhones: owning the GPU design lets a manufacturer match silicon to software in ways off-the-shelf solutions often can’t. The result could be better power efficiency for AI workloads, smoother gaming, and tighter integration with Samsung’s own ecosystem of apps and features.

What to expect and when
South Korean outlet Hankyung reports that Samsung plans to debut its in-house GPU in the Exynos 2800, slated for 2027, and to equip the Galaxy S28 series with this new architecture. If accurate, the move marks a major milestone: Samsung will transition away from AMD’s RDNA-based designs and push a proprietary graphics IP into flagship phones.
That’s a multi-year roadmap: partnerships, prototypes, and rigorous validation are required before a mobile GPU can meet the power and thermal constraints of flagship handsets. Still, Samsung’s semiconductor arm has ramped up R&D and appears determined to control more of the silicon stack.
Ambitions beyond phones
Samsung’s GPU plans aren’t limited to smartphones. The company reportedly aims to extend this graphics technology to AR glasses, in-car infotainment systems, autonomous driving platforms, and even humanoid robots. A unified, in-house GPU architecture could simplify porting graphics and AI features across a variety of devices — and create new differentiation in crowded markets.
Why does that matter? Because GPUs today do more than render pixels. They accelerate neural networks, sensor fusion, and real-time compute tasks that power AR experiences, driver-assistance systems, and interactive robotics. Owning the GPU design gives Samsung the flexibility to prioritize those workloads.
There are risks, of course. Designing a competitive GPU from scratch is costly, technically challenging, and takes time. Yet if Samsung can match or exceed the performance and efficiency of AMD’s mobile RDNA implementations while enabling deeper AI integration, it could reshape how the company positions Exynos chips across devices.
For now, the industry will be watching how Samsung balances ambition with execution. Will a custom Exynos GPU deliver a tangible edge in AI and graphics? Expect early clues in the years leading up to 2027, but the full picture may only emerge once those silicon and software optimizations ship in a flagship handset.
Comments
Daniel
I used to tweak mobile drivers, this is no small feat. Tons of validation, weird bugs, vendor politics. If Samsungs team is legit they might pull it off, but i'm skeptical.
mechbyte
Is Samsung really gonna pull off a custom GPU? Feels risky, huge R&D, thermal headaches, driver mess. If they nail AI perf and battery tho, it's a game changer.
Leave a Comment