Explain It in 30 Seconds
A console company needs a machine with a chosen balance of computing, graphics, memory, power and cost.
AMD contributes reusable technology and chip-design expertise. The platform owner contributes requirements, its own engineering and the broader hardware/software business. The result is customer-specific silicon inside a product that AMD does not own.
Where It Sits in AMD
Semi-custom products and development services contribute to the Client and Gaming reportable segment. Within it, Gaming includes both semi-custom activity and Radeon gaming graphics. [S1]
The architecture can draw on technology also used in mainstream CPUs and GPUs. The commercial arrangement differs from selling a standalone processor into an open PC channel: the chip is tied to a customer's particular platform and production schedule.
The Problem It Solved
A standard chip may not meet the combination of requirements a platform owner needs. Designing every part of a complex processor from scratch would require another large engineering investment.
Semi-custom development offers a middle route. The parties reuse appropriate supplier capabilities while making choices for the customer's system. The value comes from the balanced finished design, not from making every part unique.
How It Works — Simple Version
The customer defines the intended platform and works with AMD on the silicon requirements. The teams develop and validate the design, including its interaction with software and the rest of the hardware.
Manufacturing partners then produce the silicon, and shipments support the customer's system production. Sales of the finished platform influence later orders. Winning the design is the beginning of a product cycle, not the receipt of all its future revenue at once.
Consoles Show the Co-Design Relationship
Sony's PlayStation 5 technical disclosure identifies a custom AMD GPU within a system that also depends on storage, custom input/output and software. Microsoft's Xbox Series X account describes joint work around a custom Zen 2 CPU and RDNA 2-class GPU implementation. [S61][S62]
These disclosures establish technical relationships, not identical machines. Memory organization, interfaces, power choices and platform software affect the result. A shared architecture name cannot by itself establish that two consoles, or a console and a PC, behave the same way.
What Is Reused and What Is Customized
Reusable technology
CPU, graphics and multimedia capabilities give the design team established building blocks. Reuse can spread engineering effort across more than one commercial opportunity. [S88]
Customer requirements
The platform owner needs a chip that fits its machine, expected production economics and software plans. Spending more area or power on one function can reduce what remains for another. The design must meet the whole set of requirements, not merely win an isolated benchmark.
Platform integration
A capable chip still needs memory, storage, cooling, manufacturing test and a working software environment. Game developers then build for the platform the customer provides. The chip supplier does not replace those participants.
Development Fees and Chip Sales Are Different
AMD discloses two relevant revenue forms: non-recurring engineering fees for design/development services and sales of semi-custom SoC products to the customer. [S1]
The first concerns development work; the second concerns supplied products. Neither establishes a universal royalty on games, subscriptions or store transactions. Public reporting does not provide one per-console payment or margin that can be applied to every contract.
This distinction matters when comparing a design announcement with revenue. Engineering can be underway before production, and a shipped chip can enter a console that reaches a consumer later.
Who Uses It and Who Pays
The direct commercial customer is the platform company commissioning and purchasing the design. Its engineers are part of the development relationship.
Players use the finished console and pay the platform company, retailers or content providers for different products and services. AMD supplies silicon into that system. It does not receive the complete retail price or all spending that follows the console sale.
Why Customers Choose the Model
Co-design can let a platform owner combine differentiation with access to a supplier's established computing technology.
The trade-off includes dependence on the chosen architecture and supplier's execution. A redesign can be expensive once software and production plans are built around the platform. The relationship is deeper than a one-time spot purchase, but neither party escapes its own commercial responsibilities.
Scale and the Console Cycle
A successful platform can support substantial chip shipments over several years. Launch, growth, maturity and replacement create different demand conditions.
AMD's 2015 results illustrate that a significant semi-custom business could coexist with financial difficulty elsewhere in the company. The design wins mattered, but did not mean the whole company had already recovered. [S46]
Later cycles can move differently from other AMD markets. In Q2 2026, lower semi-custom activity weighed on Gaming while Data Center expanded. Those movements are different businesses responding to different demand, not a contradiction in one uniform AI revenue stream. [S3]
What It Depends On
The business depends on winning designs, completing them on the customer's schedule, securing production and supporting the customer's platform demand.
AMD relies on manufacturing partners; the platform also needs components AMD does not supply. Even an available processor does not guarantee that enough complete consoles can be assembled. Capacity and inventory commitments must be managed before final consumer demand is fully known. [S1]
What Depends on Semi-Custom SoCs
The selected platform depends on the custom silicon and its production continuity. Developers rely on the platform's established hardware and software behavior when creating games.
AMD, in turn, depends on a concentrated set of customer programs. A durable design win does not remove exposure to a delayed launch, a mature console cycle or a different decision for the next generation.
Why It Matters to AMD
Semi-custom provides another way to commercialize technology alongside retail Ryzen CPUs and Radeon graphics.
It can generate demand through a partner's differentiated product rather than requiring AMD to build the consumer platform itself. That expands AMD's opportunity while keeping the economics of the chip supplier separate from the platform owner's.
Before / After
The May 2013 Semi-Custom Business Unit announcement made customer-specific integration of AMD's reusable intellectual property an explicit business focus and named PlayStation 4 as its first announced design win. This was a commercialization decision, not simply another standard processor launch. [S88]
By the PS5 and Xbox Series X generation described in 2020, direct customer accounts showed continued custom-silicon relationships using later computing technology. Those later designs did not turn AMD into a console operator or establish automatic renewal of every future program. [S61][S62]
What Most People Misunderstand
A console powered by AMD is not an off-the-shelf AMD PC, and a platform's success is not all recognized as AMD revenue.
A long product cycle also contains demand risk. The customer can continue selling an established platform while ordering fewer chips, and current program success does not decide the next design competition.
Key Milestones
Evidence
Connected Entities
Ryzen and Radeon explain adjacent general-purpose product channels. Sony and Microsoft own the console platforms; AMD supplies a distinct technology and production contribution.