For Samsung, from 🤫 hussh
From the hand to the data center. The whole bill of materials.
Few companies reach as far as Samsung. The Galaxy in a person's hand. The HBM inside the AI factory. The 2nm foundry in Taylor, Texas, that prints the compute. 🤫 needs all three. A private 🤫 Agent One that lives on Galaxy, and Samsung memory and silicon powering the grid it runs on. One relationship, from the device to the data center.
One bill of materials, from Galaxy to the grid.
Galaxy at scale, with on-device Galaxy AI and the world's first 2nm smartphone chip, Exynos 2600.
🤫 Agent One runs on Galaxy as a private agent the person owns, on the Exynos NPU, by consent.
A private AI story that helps sell premium Galaxy and XR.
AI memory is driving record results, and you are growing 2nm foundry orders in Taylor, Texas.
Our AI factory grid buys HBM, DDR5, enterprise SSD and US-made 2nm silicon for its compute nodes.
A new, distributed buyer for the exact capacity you are scaling.
Devices, memory, foundry and displays, all under one roof.
We need a device surface, memory and silicon, and your breadth fits all three.
One relationship across device, memory and silicon.
From chips to devices, under one roof.
Samsung runs Device Solutions, its memory and foundry business, and Device eXperience, its Galaxy and consumer business, along with displays and Harman. Its record 2025 and 2026 are an AI memory story.
About $233B, an all-time high, up 10.9%, with about ₩43.6T of operating profit. AI memory led the way.
Device Solutions drove about 94% of operating profit in a record Q1 2026. Memory is the engine.
In mass production and designed for NVIDIA's Vera Rubin platform. The memory behind the AI factory.
The $37B Taylor fab begins operations in the second half of 2026, and Samsung plans to grow 2nm orders. Silicon made in America.
The world's first 2nm smartphone chip, on Samsung's own SF2 process, inside Galaxy S26.
"The most intuitive Galaxy AI phone yet." A device the person owns, where a private agent runs.
Samsung's reach from chips to devices is rare. It is what makes one bill of materials possible.
Memory leadership
Top-tier DRAM and NAND, with HBM4 in mass production. The memory that makes the AI factory possible.
Foundry
The world's #2 logic foundry, running 2nm gate-all-around and building a $37B fab in Texas.
Galaxy at scale
One of the two largest smartphone makers in the world, with the full Galaxy line and Android XR.
Galaxy AI on the device
Galaxy AI on the Exynos 2600 2nm NPU. Intelligence that runs locally, under the person's consent.
Breadth
Memory, foundry, displays, components, devices and Harman. A reach from chips to devices that few can match.
Enterprise SSD and DDR5
Server DDR5, enterprise SSDs and PCIe Gen6 storage. The supply layer for compute nodes.
Where Samsung is going.
AI memory leadership
HBM4, HBM4E and high-value AI products as the growth engine, with momentum "led by semiconductors."
Foundry momentum
More 2nm orders and the Taylor fab coming online. Leading-edge process, competitive again.
Galaxy AI on the device
Galaxy AI as the mobile differentiator, on the Exynos 2600 2nm NPU.
XR as a platform
Android XR with Google. A new spatial computing platform, and a new home for agents.
The process roadmap
SF2 in production, SF2P ramping, and 1.4nm targeted by 2029. The leading edge, made in America.
Galaxy S26, HBM4, and 2nm in Taylor.
The newest from Samsung, and where Agent One and the grid connect.
Galaxy S26 and Exynos 2600
On sale since March 2026, with new Galaxy AI and the first 2nm smartphone chip. The phone as a private AI device.
HBM4 for AI
In mass production and designed for NVIDIA's Vera Rubin, with HBM4E shown at GTC 2026. The AI memory story.
Galaxy XR
Samsung's first Android XR headset, built with Google and Qualcomm. A new spatial surface for a person's own agent.
2nm in Taylor, Texas
The $37B fab begins operations in the second half of 2026, with Tesla's AI6 as an anchor customer. American silicon capacity.
Buy together. Build together. Sell together. Take risk together.
We buy Samsung memory and silicon for the grid, build 🤫 Agent One into Galaxy, and tell a private AI story together.
Build together: Agent One on Galaxy
Galaxy S26, foldables and XR, with on-device Galaxy AI and the Exynos 2600 NPU.
🤫 Private Agent One as a first-class experience on Galaxy, running locally on the Exynos NPU, by consent.
A private AI story that helps Samsung sell premium Galaxy and XR.
Sell together: the channels Samsung already has
Samsung reaches people through the Galaxy Store, Knox device management and carriers worldwide.
🤫 Agent One as a consent-first private AI layer on Galaxy: in the Galaxy Store, managed through Knox for enterprise and government fleets, and offered by carriers.
Samsung sets Galaxy apart, and every person gets a private agent through a channel they already trust.
Buy together: memory and foundry for the grid
Samsung is growing HBM and 2nm foundry demand.
🤫 Puppy One and grid nodes buy DRAM, enterprise SSD, HBM4 and HBM4E, and US-made 2nm silicon from Taylor.
A new, distributed buyer for the capacity Samsung is scaling, made in America.
Take risk together: the bold bets
Samsung is investing in on-device AI silicon, Android XR and edge memory.
We co-invest in private agent silicon and firmware, XR as an agent surface, and memory for edge supercomputing.
Shared risk on private, human-controlled AI infrastructure.
The people building Samsung.
TM Roh
Co-CEO, Device eXperience (DX)
Leads Galaxy, XR and consumer devices, the experiences where a private agent lives.
Young Hyun Jun
Vice Chairman and Co-CEO, Device Solutions (DS)
Leads memory and foundry: the AI memory engine and the 2nm supply layer.
Janghyun Yoon
President, CTO (DX) and Head of Samsung Research
Leads the device technology and on-device AI roadmap, where an Exynos-native 🤫 Agent One would run private inference on Galaxy.
From the person's device to the silicon in the grid.
Samsung reaches from the device where 🤫 Agent One lives to the memory and foundry that power the grid it runs on. A fab in Taylor, Texas, makes it American, which matters to the public servants we serve. Own your AI, own your data, own your compute. And 🤫 Agent One is free to every American.