Summary: On January 5, 2026, with the CES 2026 show floor about to open, Samsung co-CEO TM Roh handed Reuters the target that defines the company’s year: shipments of Galaxy AI-equipped mobile devices will double to 800 million units, phones and tablets included. Teams building AI features for the Android ecosystem should just assume the default assistant is Gemini; integrating deeply with its APIs beats building a substitute on practicality alone.
On January 5, 2026, with the CES 2026 show floor about to open, Samsung co-CEO TM Roh handed Reuters the target that defines the company’s year: shipments of Galaxy AI-equipped mobile devices will double to 800 million units, phones and tablets included. And the engine underneath Galaxy AI is Google’s Gemini.
Anyone building AI products should pause on that figure. Eight hundred million units sits in the neighborhood of a full year’s shipments for a top-tier phone brand. Samsung is effectively announcing that in 2026, an “AI device” stops being flagship marketing copy and becomes the factory default across the whole line — and the default model is not Samsung’s own but Google’s.
The 400-to-800-Million Bet
Per Reuters, Samsung intends to grow Galaxy AI-equipped devices from roughly 400 million in 2025 to 800 million in 2026. TM Roh left zero room for hedging: “We will apply AI to all products, all functions and all services as quickly as possible.” Il Sole 24 Ore’s coverage adds that the expansion runs across Samsung’s complete product lineup, with Google celebrating the arrangement in the open.
Set against the CES 2026 timeline, two AI shipping strategies are running side by side. NVIDIA’s keynote staked its claim on physical AI — robots, autonomous vehicles, digital twins (see our NVIDIA CES 2026 physical AI coverage). Samsung staked its own on every phone becoming an AI device. One outfit sells compute; the other sells installed base. Together they define 2026 as the year of scale.
What Galaxy AI Actually Includes
Galaxy AI is not a feature but a bundle threaded through the system: Circle to Search, live call translation, a writing assistant, generative photo editing, and generative wallpapers. Coverage already spans the Galaxy S23/S24/S25 series, the FE variants, Z Fold and Z Flip 5 through 7, and the Tab S9 through S11 tablets.
For Google, the deal is about distribution at its core. Gemini graduates from “an app you install” to the default experience across 800 million devices; CNET’s same-day report flatly describes them as “powered by Google’s Gemini.” With the assistant entry point fused to the OS entry point, Google occupies the very position search-era companies fantasized about: the user installs nothing at all.
The Memory Squeeze: The Hidden Cost of 800 Million Devices
The scale pledge has a flip side, and it lives in the supply chain. CNET reports that roughly 40% of DRAM output from Samsung and SK Hynix is projected to flow to OpenAI alone; that Apple already shipped the iPhone 16 with 16GB of RAM to shoulder Apple Intelligence; and that Micron went as far as retiring Crucial, its consumer memory brand since 1996, to swing capacity toward enterprise AI customers. Data center memory demand is elbowing consumer product lines aside: analysts anticipate price increases across phones and automotive electronics, and even a potential revival of 4GB smartphones alongside graphics cards trimmed to 8-12GB of memory.
The sharpest irony lands on Samsung itself — one of the world’s largest memory producers being squeezed by the very AI demand it is fanning. On possible price hikes, TM Roh neither promised nor ruled them out, offering only: “As this situation is unprecedented, no company is immune to its impact.”
What It Means for Developers and Product Teams
First, Gemini’s distribution surface doubles in 2026. Teams building AI features for the Android ecosystem should just assume the default assistant is Gemini; integrating deeply with its APIs beats building a substitute on practicality alone. Second, AI features slide from differentiator to table stakes. Once 800 million devices wear a Galaxy AI badge, the premium for merely “having AI” caves in quickly, and the competition relocates to integration depth and real-world usefulness. Third, the hardware cost floor is lifting. Memory prices and capacity requirements push bill-of-materials costs upward, so any hardware or app business priced on 2025 cost assumptions owes itself a fresh calculation for 2026 — and if component prices climb across the board, the doubling target and the margin target will begin tugging against each other, a tension worth tracking in Samsung’s quarterly results.