Synopsys and Amazon announced a strategic, multi-year agreement on September 30, 2026, covering silicon intellectual property and engineering software for Amazon’s custom-chip development. The deal is worth more than $1 billion, according to the companies, and it marks one of the largest electronic design automation (EDA) licensing agreements tied to a single hyperscaler disclosed this year. Amazon will serve as the lead customer for Synopsys’ application-optimized IP business, a role that puts the world’s largest cloud provider at the center of how Synopsys (Nasdaq: SNPS) builds and sells chip-design blocks going forward.
The news landed the same week AI infrastructure spending kept climbing across the hyperscaler sector, and it reflects a pattern that has become familiar in 2026: cloud giants locking down the design tools and IP blocks they need to keep building their own chips rather than buying everything off the shelf from merchant silicon vendors. Coverage of the agreement has circulated through Yahoo! Finance Canada, tech-insider.org, Startup Fortune, and openPR.com, each describing the same core terms, though none have published the full contract or disclosed its exact duration.
Synopsys and Amazon’s $1 Billion-Plus Deal, Explained
At its core, the agreement is a supply arrangement for the building blocks of modern chips. Synopsys sells two broad categories of product: design IP, meaning pre-built, licensable circuit blocks that chipmakers drop into their own silicon instead of designing from scratch, and EDA software, the simulation and verification tools engineers use to lay out, test, and validate a chip before it goes to a foundry. Under this new agreement, Amazon gets access to both, with the explicit aim of accelerating its custom-silicon programs and the AI-powered products and cloud infrastructure built on top of them.
What stands out in the reporting is the “lead customer” framing. Synopsys has sold application-optimized IP, meaning IP blocks tuned for specific workloads rather than generic use, to many chip designers over the years. Naming Amazon the lead customer for that line of business signals that Amazon’s requirements, and its purchase volume, will now shape how Synopsys prioritizes and develops those IP blocks. That is a meaningfully different relationship than a standard licensing contract, and it is why the deal registered as a governance and strategy story in the semiconductor press rather than a routine vendor renewal.
Table: The Synopsys-Amazon Deal at a Glance
| Detail | What’s Confirmed |
|---|---|
| Announcement date | September 30, 2026 |
| Companies involved | Synopsys, Inc. (Nasdaq: SNPS) and Amazon |
| Agreement type | Strategic, multi-year agreement |
| Disclosed value | More than $1 billion |
| Amazon’s role | Lead customer for Synopsys’ application-optimized IP business |
| Commercial model | Shift toward license-plus-royalty structure |
| Technologies covered | EDA, simulation and analysis (S&A), agentic AI technologies |
| Stated purpose | Accelerate Amazon’s custom-silicon development and AI infrastructure |
| Exact contract length | Not disclosed |
| Specific IP blocks or chip designs | Not disclosed; companies did not detail which products are covered |
Why Amazon Is Betting Bigger on Custom Silicon
Amazon has spent nearly a decade building an internal chip-design practice through its Annapurna Labs unit, and this agreement extends that bet rather than starting a new one. The company has already shipped multiple generations of Arm-based server processors under the Graviton name, AI training and inference silicon under the Trainium name, and networking and virtualization offload chips under the Nitro name. Those three product lines were specifically named in reporting on this deal as part of Amazon’s custom-chip portfolio that stands to benefit from deeper access to Synopsys IP and tooling.
The logic is straightforward even if the engineering is not. Every cloud provider racing to field more AI compute is constrained by two things: how much merchant silicon they can buy from Nvidia and AMD, and how fast their own design teams can turn new chips into production hardware. Expanding a silicon IP and EDA relationship with Synopsys addresses the second constraint directly. More and better IP blocks, paired with simulation and validation software tuned for Amazon’s own workflows, should in theory cut the time between a chip concept and a working data center part.
Amazon’s Named Custom Silicon Lines
| Chip Line | General Role |
|---|---|
| Nitro | Virtualization, networking, and security offload for EC2 instances |
| Graviton | Arm-based general-purpose server processors for EC2 |
| Trainium | AI training and inference accelerators, an alternative to GPU instances |
Amazon’s cloud arm already runs these chips alongside Amazon EC2 instances powered by merchant GPUs, giving customers a choice between Nvidia-class accelerators and Amazon’s own silicon depending on price and workload. A deeper IP relationship with Synopsys does not replace that merchant-silicon business, but it does give Amazon more room to expand the in-house side of the ledger.
Inside Synopsys’ Application-Optimized IP Business
Synopsys built its reputation on EDA, the category of software that chip designers use to draw, simulate, and verify a design before committing it to a foundry. Over the past several years the company has pushed further into design IP itself, selling pre-verified blocks such as interface controllers, memory interfaces, and processor cores that customers integrate directly into their chips instead of designing equivalent logic from scratch.
Application-optimized IP takes that one step further: instead of a generic, one-size-fits-all block, the IP is tuned for a specific class of workload, such as AI training, high-bandwidth memory access, or network switching. For a company like Amazon, designing chips for the narrow, demanding task of running AI models at cloud scale, that specialization matters more than it would for a general-purpose chipmaker. Naming Amazon as the lead customer for this business line suggests Synopsys expects AI-focused hyperscalers to be the biggest growth driver for application-optimized IP over the next several years.
The License-Plus-Royalty Shift Explained
One of the more specific details in the reporting is that Synopsys’ silicon-IP business is expanding toward a license-plus-royalty commercial model. Traditional IP licensing in the semiconductor industry typically involves a flat or tiered license fee paid up front, with the customer free to use the IP as many times as it wants within the terms of the contract. A license-plus-royalty structure adds a second layer: Synopsys collects an ongoing royalty tied to how much the licensed IP is actually used, often measured per chip shipped or per unit sold.
For Synopsys, that model aligns revenue with customer success. The more chips Amazon ships using Synopsys IP, the more Synopsys earns, rather than collecting a single payment regardless of production volume. For Amazon, it likely means a lower upfront cost in exchange for payments that scale with deployment, a tradeoff large chip buyers often accept when they expect high-volume, long-running production rather than a one-off design. Neither company has disclosed royalty rates or the specific IP categories the model applies to, so the practical financial impact on either side stays undisclosed for now.
EDA, Simulation and Analysis, and Agentic AI: What Synopsys Brings
Beyond IP blocks, the agreement covers three categories of Synopsys technology: electronic design automation, simulation and analysis software, and what the companies describe as agentic AI technologies. The first two are the backbone of Synopsys’ decades-old core business, used by chip designers across the industry to lay out circuits, run timing analysis, and catch defects before fabrication. The third category is newer and reflects where the whole EDA industry has been heading for the past two years: using AI agents to automate parts of the chip-design workflow that previously required teams of engineers working through repetitive verification cycles by hand.
The companies describe plans to apply AI across what they call silicon-to-system engineering workflows, plus joint work on custom AI capabilities for chip design and validation specifically. That framing matters because it suggests this is not simply Amazon buying off-the-shelf Synopsys tools. It points toward Synopsys and Amazon co-developing AI-driven design capabilities, with Amazon’s cloud infrastructure providing the compute and Synopsys providing the domain-specific chip-design expertise.
How AWS Services Power the Partnership
Reporting on the deal names three specific AWS services Synopsys plans to use for its own product development and AI applications going forward: Amazon Elastic Compute Cloud (EC2), AWS cloud-storage services, and Amazon Bedrock. That detail flips the usual vendor-customer script. Synopsys is not just selling IP and software to Amazon, it is also becoming a customer of AWS infrastructure to build and run the very AI capabilities it is developing under this agreement.
Running EDA workloads on cloud infrastructure is not new. Chip simulation and verification runs are notoriously compute-hungry, and EDA vendors have spent years moving those workloads to the cloud to avoid the cost of maintaining on-premises server farms that mostly sit idle outside of crunch periods. What is new here is the explicit tie-in to Bedrock, Amazon’s managed service for building generative AI applications, which points toward Synopsys layering large language model-based tools on top of its simulation and design software rather than treating cloud compute purely as a capacity backstop.
Historical Context: Synopsys’ Long Run as an EDA Powerhouse
Synopsys has operated as one of the handful of companies that effectively every chipmaker on Earth depends on, whether that chipmaker ever mentions the name publicly. Along with Cadence Design Systems and Siemens EDA (formerly Mentor Graphics), Synopsys forms what the industry has long called the “Big Three” of electronic design automation. Almost no modern chip, from a smartphone processor to a data center accelerator, reaches production without passing through tools built by one of these three companies at some stage of its design cycle.
What has shifted over the past several years is the customer base. EDA vendors built their businesses selling to traditional chipmakers: Intel, Qualcomm, Broadcom, and the like. The rise of hyperscaler custom silicon changed that. Amazon, Google, and Microsoft each now run internal chip-design teams building processors and AI accelerators for their own data centers, and each needs the same category of EDA tools and IP that traditional chipmakers have relied on for decades. This Amazon deal is the clearest sign yet that Synopsys now treats hyperscalers as first-tier customers on par with its historic semiconductor clients, not as a side business.
Competitive Landscape: Synopsys vs Cadence in Chip Design Software
Cadence Design Systems remains Synopsys’ closest rival across both EDA tools and design IP, and the two companies have competed for hyperscaler business for years as cloud providers built out internal silicon teams. Landing Amazon as a named lead customer for application-optimized IP gives Synopsys a visible marker of how deep that relationship runs, something that is harder to measure from the outside when deals stay private, as most EDA-hyperscaler agreements historically have.
| Factor | Synopsys | Cadence |
|---|---|---|
| Exchange / ticker | Nasdaq: SNPS | Nasdaq: CDNS |
| Core business | EDA software, silicon IP, simulation and analysis | EDA software, silicon IP, system analysis |
| Confirmed hyperscaler mega-deal (2026) | Amazon, $1B+, announced Sept. 30, 2026 | No comparable disclosed deal in this report |
| Reported new focus area | Agentic AI technologies in design workflows | AI-driven verification and design tools (industry-wide trend) |
It is worth being precise about what this table does and does not show. It is not a claim that Cadence lacks its own hyperscaler relationships. It reflects what has been publicly confirmed about this specific Synopsys-Amazon agreement versus the absence of an equivalent disclosed deal from Cadence in the same reporting cycle. The broader EDA competitive dynamic between the two companies continues across many other fronts, including traditional semiconductor customers that this deal does not touch.
Market Impact: What This Means for the EDA Sector
Deals of this size send a signal beyond the two companies involved. When a hyperscaler the size of Amazon commits to a nine-figure-plus, multi-year IP and software agreement, it tells the rest of the chip-design industry that custom silicon is not a side project at AWS, it is a long-term capital commitment backed by board-level contracts. That matters for every smaller chip-design shop and fabless startup competing for Synopsys’ and Cadence’s engineering attention, since large anchor customers tend to shape product roadmaps for years at a time.
It also reinforces a trend that has been building across the AI infrastructure buildout in 2026: the money flowing into AI is not just going to GPU makers and cloud capacity. A meaningful share is going to the design tools and intellectual property layer that sits underneath every custom chip a hyperscaler ships, a layer that rarely gets headline attention compared to GPU shipment numbers or data center capex figures, but one this deal shows is scaling right alongside them.
What This Means for AWS’s Custom Silicon Strategy Going Forward
For AWS specifically, the agreement strengthens a strategy the company has pursued since it acquired Annapurna Labs in 2015: reducing dependence on any single external chip supplier by building credible in-house alternatives. Graviton has already demonstrated that an Arm-based cloud processor line can reach broad customer adoption. Trainium is AWS’s answer to the GPU shortage and pricing pressure that has defined much of the AI infrastructure market through 2026. Deeper Synopsys IP access should, in principle, shorten the design cycle for whatever comes after the current Trainium and Graviton generations.
The agentic AI component of the deal is worth watching closely over the next year. If Synopsys and Amazon do manage to apply AI agents meaningfully across silicon-to-system engineering workflows, as the companies have stated they intend to, it could compress Amazon’s chip-design timelines in ways that are hard to achieve through IP licensing alone. That is the part of this agreement with the most upside, and also the part with the least public detail so far.
Predictions: Where This Partnership Goes From Here
Based on the confirmed terms of the deal and the broader direction of the hyperscaler custom-silicon market, a few outcomes look likely over the next 12 to 18 months, though none of these are confirmed by the companies and should be read as analysis rather than fact.
- Expect Synopsys to disclose more hyperscaler-scale IP deals in coming quarters, following the same license-plus-royalty pattern it is applying to Amazon, as Google and Microsoft push their own custom-silicon programs further.
- Expect AWS to introduce a new Trainium or Graviton generation that references expanded IP access or faster design cycles, even if Synopsys is not named directly in product marketing.
- Expect Cadence to respond with its own hyperscaler-anchored announcement, given how directly this deal positions Synopsys ahead in the AI-chip IP conversation.
- Expect agentic AI design tools to become a standard talking point across EDA vendor earnings calls through 2027, following the direction both Synopsys and Cadence have signaled this year.
- Expect scrutiny of royalty-based IP licensing to grow among chip-design customers, as more hyperscalers weigh the tradeoff between lower upfront costs and long-term royalty exposure at high production volumes.
What’s Still Unknown About the Agreement
For all the detail that has emerged, several material facts remain undisclosed. The companies have not specified the agreement’s exact duration, nor have they detailed which specific Synopsys IP blocks or which Amazon chip designs the agreement actually covers. No on-record executive quote from either company has circulated alongside the announcement in the reporting reviewed for this story. That gap is common for deals of this size in their first days of coverage, and further detail typically surfaces in subsequent earnings calls, where both Synopsys and Amazon disclose material contracts to investors under standard reporting obligations.
Readers tracking the semiconductor supply chain more broadly, including industry-wide trends tracked by the Semiconductor Industry Association, should expect this deal to come up as a reference point in discussions about hyperscaler chip strategy well into 2027, given how directly it ties Synopsys’ IP roadmap to Amazon’s AI infrastructure buildout.
FAQ: Synopsys and Amazon’s Silicon IP Deal
What did Synopsys and Amazon actually announce?
A strategic, multi-year agreement, announced September 30, 2026, covering silicon intellectual property and engineering software for Amazon’s custom-chip development. The deal is worth more than $1 billion.
How much is the Synopsys-Amazon deal worth?
The companies have disclosed the value as more than $1 billion. The exact total contract value and duration have not been made public.
What is application-optimized IP?
It refers to pre-built chip-design blocks tuned for specific workloads, such as AI training or high-speed memory access, rather than generic, one-size-fits-all circuit blocks. Amazon is now the lead customer for this line of Synopsys’ business.
What is a license-plus-royalty model in chip IP licensing?
It is a commercial structure combining an upfront license fee with ongoing royalty payments tied to usage, often measured per chip shipped, rather than a single flat license fee covering unlimited use.
Which Amazon chips does this deal affect?
Reporting names Nitro, Graviton, and Trainium as part of Amazon’s custom-chip portfolio in connection with the deal. The companies have not disclosed exactly which specific designs will incorporate the new Synopsys IP.
Which AWS services will Synopsys use under this agreement?
Reports name Amazon EC2, AWS cloud-storage services, and Amazon Bedrock as services Synopsys will use for product development and AI applications.
Is this Synopsys’ biggest hyperscaler deal to date?
It is the largest publicly disclosed hyperscaler IP and software agreement Synopsys has announced in 2026 based on currently available reporting, though the companies have not characterized it as the company’s largest deal overall.
Did Synopsys or Amazon provide an executive quote about the deal?
No verifiable, named executive quote has circulated alongside the announcement in the reporting reviewed for this story. Any statements attributed to specific executives should be treated cautiously until the companies publish an official statement with named quotes.
