A multi-year, $1B+ agreement between Synopsys and Amazon introduces a royalty leg that ties the IP supplier’s revenue to the customer’s chip output — and runs a reverse leg in the opposite direction.
Every figure below is Synopsys’ own characterisation, from its news release of 30 September 2026. The release gives no term length, no royalty rate and no minimum commitment, and nothing below estimates them. This is a commercial agreement, not a merger, acquisition or equity investment, and nothing here takes a view on whether the structure favours either company. Nothing here is investment advice.
What Happened
Synopsys and Amazon announced a strategic, multi-year agreement on 30 September 2026. The announcement is datelined Sunnyvale, California, and was distributed through PR Newswire. Synopsys trades on Nasdaq under the ticker SNPS.
The two companies describe the agreement as building on more than 15 years of collaboration. It expands Amazon’s use of Synopsys IP to include application-optimised silicon IP, alongside its EDA, simulation and analysis, and agentic AI technologies.
The agreement expands Synopsys’s silicon IP business to serve what both companies describe as increasing demand for application-optimised IP. The release names Amazon as Synopsys’s lead customer for that IP category. Three Amazon chip families are explicitly covered: Nitro, which handles cloud security, network and storage; Graviton, used for general-purpose computing; and Trainium, built for AI training and inference workloads.
The agreement also expands Amazon’s use of Synopsys EDA, simulation and analysis, and agentic AI technologies. The two companies describe joint work to accelerate multiphysics solutions on Trainium and Graviton, and plan custom agentic AI capabilities to help Amazon’s engineering teams design, analyse, optimise and validate chips and systems. More than fifteen years of prior collaboration between the two companies is cited in the release.
The key insight: The model change — not the headline figure — is the structural event. Under a flat licence, Synopsys is paid identically whether Amazon ships ten thousand chips or ten million. Under licence-plus-royalty, the realised total moves with production volumes. The release states the model “aims to deliver value to both companies as production volumes grow.” That is upside participation for Synopsys. It is also, necessarily, exposure to volume outcomes that have not yet occurred and that the release does not disclose.
The Structural Read
The release’s own subhead names the structural shift: the deal “marks the expansion of Synopsys’ silicon IP business to serve increasing demand for application-optimized IP while evolving to a license-plus-royalty business model that aims to deliver value to both companies as production volumes grow.”
Under a flat-licence model, the IP vendor receives a fixed fee. Volume risk sits entirely with the customer. Volume upside stays entirely with the customer too.
A royalty changes the distribution in both directions. Synopsys now participates in the outcome of Amazon’s chip output — upward if volumes grow, and exposed if they do not. The release gives no royalty rate, no minimum commitment, no term in years, and no annual split. The $1B+ figure is the company’s own characterisation. It is not a filed number and it is not broken down. With a royalty component, what the agreement ultimately realises depends on production volumes that have not been disclosed.
The second structural feature runs in the opposite direction. Synopsys says it is adopting Amazon EC2, Amazon cloud storage, and Amazon Bedrock to build and deploy AI applications and agents for its own product development. So Synopsys sells IP to Amazon and buys compute from AWS — in the same announcement.
That two-way structure is the half most coverage drops. Most reporting treats the $1B+ figure as the story. The reverse leg — an IP vendor publicly committing to the cloud and AI infrastructure of its largest IP customer — is a disclosed commercial structure worth noting separately from the headline number.
Peter DeSantis, SVP Foundational AI, Custom Silicon & Quantum Computing, Amazon
“Purpose-built chips deliver better performance at lower cost because they’re designed for exactly what customers need.”
Sassine Ghazi, President and CEO, Synopsys
“We are proud to have Amazon as our lead customer for this next phase of growth.”
What the Release Does and Does Not Establish
Concrete items from the release: Amazon is named as lead customer for application-optimised IP. The agreement expands Amazon’s use of Synopsys EDA, simulation and analysis, and agentic AI technologies. Synopsys is adopting Amazon EC2, cloud storage, and Amazon Bedrock. The two companies plan custom agentic AI capabilities for Amazon’s chip engineering teams. The agreement is characterised as $1B+ and multi-year.
Not established by anything in the release, and therefore absent here: the term of the agreement in years, the royalty rate, any minimum volume commitment, revenue-recognition timing, which specific IP blocks are licensed, and what Synopsys pays AWS for the compute it is adopting. This is a commercial agreement. It is not a merger, an acquisition, a partnership with an equity stake, or an investment.
Three Implications
THE MODEL SHIFT IS THE UNIT OF ANALYSIS
A flat licence and a licence-plus-royalty are structurally different instruments. Under the flat model, volume risk and volume upside belong to the customer. Under the royalty model, both move partly to the supplier. The release describes this shift explicitly. Analysts and readers tracking Synopsys’s revenue profile have a new variable: Amazon’s chip output volumes, which are not disclosed here or elsewhere in the release.
THE REVERSE LEG CREATES A MUTUAL DEPENDENCY
Synopsys sells silicon IP to Amazon. Synopsys also buys compute — EC2, cloud storage, Bedrock — from Amazon. Each firm is supplier and customer to the other simultaneously. For IP vendors whose customers are increasingly designing their own silicon rather than buying standard chips, this kind of interlocking commercial structure is a disclosed data point about how those relationships can be organised.
THE $1B+ FIGURE IS A FLOOR-SHAPED HEADLINE
The $1B+ is the company’s own characterisation. It is not a filed figure. It carries no breakdown, no annual split, and no disclosed royalty rate. Because a royalty component is present, the realised total depends on production volumes that have not happened yet. Treating the number as a ceiling, a floor, or a mid-point is not supported by the release. It is a headline attached to an outcome that is, in part, still contingent.
