Samsung’s next Exynos chip could loosen Qualcomm’s grip on Galaxy phones, provided its foundry stops tripping over yields.
Samsung made history by unveiling its first 2nm gate-all-around chipset, the Exynos 2600. Now the Korean outfit is shifting engineers and cash towards the Exynos 2700.
The next system-on-chip is expected to use an improved version of Samsung’s 2nm node. It should appear just before the Galaxy S27 series lands in early 2027.
Rivals might dismiss the Exynos 2600 as a one-off bit of corporate chest-thumping. The Exynos 2700 looks more serious, and that could make Qualcomm and TSMC twitchy. For Qualcomm, Samsung’s home-grown silicon means fewer Snapdragon sales. For TSMC, it is a reminder that Samsung has not given up trying to claw back foundry business.
Rising 2nm wafer costs and DRAM prices are making life harder for phone makers. Margins get squeezed, and suddenly Samsung’s own foundry starts looking less like a vanity project.
The dreaded Qualcomm tax is the bit Samsung wants to dodge. The Snapdragon 8 Elite Gen 6 Pro is estimated to cost more than $300 per unit, which is not exactly pocket fluff. That makes the Exynos 2700 attractive, even if it does not beat Qualcomm’s chip outright. A slightly slower but more efficient part could still save Samsung a pile of cash.
Samsung is reportedly planning to use the Exynos 2700 in 50 per cent of Galaxy S27 shipments. That would be a big jump from the Exynos 2600, which is expected in 25 per cent of Galaxy S26 units.
Qualcomm will not enjoy that maths. Its handset business is already feeling the squeeze from the memory shortage, forcing it to hunt for growth elsewhere. Samsung ships more Android flagships than anyone else. Losing half of that Galaxy S27 socket to Exynos would sting, even for Qualcomm.
TSMC has a different worry. The Exynos 2700 would show that Samsung is still inching towards relevance in advanced foundry work. TSMC dominates because it spends big on advanced nodes and gets stable yields. Customers want wafers they can actually use, not press releases wrapped in silicon.
Samsung’s problem has always been yield. Being first with flash process technology means less when too many chips come out of the fab looking like expensive coasters. Reports suggest Samsung’s 2nm GAA yields have not yet moved past 60 per cent. To tempt customers away from TSMC, that figure probably needs to reach at least 70 per cent.
That is why Samsung is not yet seen as a proper alternative, even with TSMC drowning in AI-driven wafer demand. Customers may look at Samsung’s lithography, but they still count usable chips.
The Exynos 2800 is reportedly staying on Samsung’s 2nm GAA process. That suggests Samsung may care more about optimisation than racing TSMC to the next process badge. That is probably sensible. Samsung does not need to win the nanometre naming contest if it can make chips that are fast, efficient and cheap enough.
