Amazon signs $1 billion Synopsys deal for AWS chips

Amazon signs $1 billion Synopsys deal for AWS chips: the analysis positions SNPS, AMZN and TSM to benefit and INTC, NVDA and AMD under pressure; confidence high.

Analyzed 2026-10-01 — point-in-time, never revised. Outcomes are scored against SPY and published either way.

Ranked exposures

  1. ▲ SNPS Synopsys — impact +85 · confidence high · horizon immediate · 1st order
    Synopsys secures over one billion dollars in multiyear licensing revenue from Amazon's custom AWS silicon programs.
    Chain: Amazon signs multiyear IP agreement with Synopsys → Synopsys books over one billion in licensing commitments → IP and EDA cloud revenue backlog rises → Margin mix improves on high-margin IP royalties
    What would break the call: Synopsys discloses the billion dollars spreads over many years with no backlog uplift in quarterly reporting.
  2. ▲ AMZN Amazon — impact +60 · confidence high · horizon quarters · 1st order
    Amazon lowers AWS compute costs by deploying in-house accelerators that reduce capital spending per training hour.
    Chain: Amazon licenses broad IP portfolio for custom silicon → Internal Trainium and Graviton design cycles accelerate → Cost per training hour falls versus merchant GPUs → AWS operating margin and capex efficiency improve
    What would break the call: Trainium adoption stalls and AWS keeps buying merchant GPUs at scale, leaving capex per workload flat.
  3. ▲ TSM TSMC — impact +50 · confidence high · horizon quarters · 2nd order
    TSMC gains advanced-node manufacturing revenue as Amazon moves additional Trainium and Graviton designs into production.
    Chain: Amazon expands internal chip design programs → More tapeouts move to advanced nodes → TSMC wins wafer and CoWoS packaging volume → Advanced node revenue mix rises
    What would break the call: Amazon dual sources leading-edge production to Intel Foundry or Samsung, diverting wafer volume away.
  4. ▲ ARM Arm — impact +48 · confidence high · horizon quarters · 2nd order
    Arm earns higher licensing revenue and royalties as Amazon deploys more Graviton processors across expanding AWS infrastructure.
    Chain: Amazon accelerates Graviton design with licensed IP → Graviton deployments expand across AWS regions → Arm collects per chip royalties on shipments → Royalty revenue from infrastructure compute rises
    What would break the call: Amazon shifts future server cores to RISC-V, cutting Arm royalty per deployed socket.
  5. ▼ INTC Intel — impact -45 · confidence high · horizon quarters · 2nd order
    Intel loses AWS server processor sales as Amazon expands Graviton deployments and displaces x86 sockets.
    Chain: Amazon speeds internal Arm server CPU development → Graviton share of AWS instances increases → x86 socket purchases at AWS decline → Intel data center group revenue falls
    What would break the call: Amazon awards Intel Foundry a Trainium or Graviton manufacturing contract that offsets lost socket revenue.
  6. ▼ NVDA Nvidia — impact -45 · confidence high · horizon quarters · 2nd order
    Nvidia loses AWS accelerator sales as Amazon shifts training workloads toward internally designed Trainium processors.
    Chain: Amazon shortens Trainium design cycles with licensed IP → Internal accelerator supply grows at AWS → Share of AWS training workloads shifts internal → Nvidia unit sales into AWS grow slower
    What would break the call: AWS announces another multi gigawatt Nvidia Blackwell or Rubin buildout, confirming merchant GPU demand holds.
  7. ▲ CDNS Cadence Design Systems — impact +42 · confidence medium · horizon weeks · 3rd order
    Cadence sells more simulation and verification tools as hyperscalers expand custom silicon budgets following Amazon's Synopsys commitment.
    Chain: Amazon deal validates hyperscaler custom silicon spending → Rival hyperscalers increase their own design budgets → EDA and verification tool demand broadens → Cadence bookings and backlog rise
    What would break the call: Synopsys wins exclusive hyperscaler IP agreements that lock Cadence out of comparable accounts.
  8. ▼ AMD AMD — impact -40 · confidence high · horizon quarters · 2nd order
    AMD loses AWS processor and accelerator sales as Amazon replaces merchant silicon with internally designed Graviton and Trainium chips.
    Chain: Amazon strengthens internal CPU and accelerator roadmap → AWS reduces merchant silicon purchase intent → AMD Epyc and Instinct volumes into AWS shrink → Data center segment growth decelerates
    What would break the call: AWS announces a large Instinct MI400 deployment, showing merchant accelerators retain share despite Trainium.
  9. ▲ ALAB Astera Labs — impact +40 · confidence low · horizon quarters · 2nd order
    Custom accelerator racks need connectivity silicon, raising Astera Labs shipments of retimers and fabric switches into AWS systems.
    Chain: Amazon builds more custom accelerator racks → Rack scale designs need PCIe and CXL connectivity → Astera retimer and fabric content per rack rises → Astera shipment volume into AWS grows
    What would break the call: Amazon integrates retimer and fabric functions into its own silicon, removing Astera content per rack.
  10. ▲ MU Micron Technology — impact +40 · confidence low · horizon quarters · 2nd order
    Trainium accelerators consume high bandwidth memory stacks, adding Micron shipment volume as Amazon expands internal silicon production.
    Chain: Trainium volumes expand on faster design cycles → Each accelerator pairs with HBM stacks → Memory bit demand from AWS rises → Micron HBM revenue and pricing firm
    What would break the call: Amazon qualifies only SK Hynix and Samsung HBM for Trainium, leaving Micron without allocation.
  11. ▲ AMKR Amkor Technology — impact +35 · confidence low · horizon quarters · 3rd order
    More custom accelerator tapeouts raise advanced packaging and test volume that Amkor services for hyperscaler silicon.
    Chain: Amazon increases custom chip tapeout cadence → Advanced packaging and test demand rises → Amkor books higher OSAT volume → Communications and computing segment revenue grows
    What would break the call: TSMC keeps all Amazon advanced packaging in house at CoWoS lines, bypassing external OSAT partners.
  12. ▲ CRDO Credo Technology — impact +35 · confidence low · horizon quarters · 2nd order
    Scaling Trainium clusters lifts demand for active electrical cables and SerDes chiplets that Credo supplies to hyperscalers.
    Chain: Amazon scales internal accelerator cluster deployments → Cluster scale-up needs high speed interconnect → Credo AEC and SerDes shipments to AWS rise → Revenue concentration from hyperscalers increases
    What would break the call: AWS standardizes on optical interconnect for scale-up links, displacing Credo copper cable attach.
  13. ▼ MRVL Marvell Technology — impact -31 · confidence medium · horizon weeks · 2nd order
    Marvell loses AWS custom ASIC opportunities as Amazon prioritizes internally designed silicon supported by Synopsys.
    Chain: Amazon licenses IP to build silicon internally → Need for external ASIC design partner falls → Marvell custom compute pipeline at AWS narrows → Custom silicon revenue guidance softens
    What would break the call: Marvell announces a new multi generation AWS custom silicon win covering Trainium or networking.
  14. ▼ ALGM Alphawave IP — impact -30 · confidence low · horizon quarters · 2nd order
    Rival connectivity IP vendor loses share as Amazon standardizes on Synopsys interface IP for custom silicon.
    Chain: Amazon consolidates interface IP purchases with Synopsys → Competing SerDes IP vendors lose hyperscaler sockets → Alphawave design win pipeline narrows → Licensing and royalty revenue growth slows
    What would break the call: Amazon dual sources SerDes IP for specific Trainium generations, preserving Alphawave design wins.
  15. ▲ ANET Arista Networks — impact +30 · confidence low · horizon quarters · 2nd order
    Larger internal accelerator clusters need more Ethernet switching ports, lifting Arista cloud titan networking revenue.
    Chain: Amazon scales custom accelerator cluster deployments → Scale-out fabrics require more Ethernet switch ports → Arista cloud titan orders increase → AI networking revenue guidance rises
    What would break the call: Amazon builds its own switching silicon and white box fabric, displacing Arista from back end networks.
  16. ▼ CRWV CoreWeave — impact -25 · confidence low · horizon quarters · 3rd order
    Lower cost Amazon silicon pressures rented GPU pricing, squeezing CoreWeave contract rates as cloud capacity grows.
    Chain: Cheaper internal AWS silicon lowers compute cost → AWS prices accelerator instances more aggressively → Rented GPU market pricing resets lower → CoreWeave contract rates and margins compress
    What would break the call: AI compute demand stays supply constrained, keeping CoreWeave contract pricing and utilization firm.
  17. ▼ TXN Texas Instruments — impact -25 · confidence low · horizon quarters · 3rd order
    Texas Instruments analog chip volume drops after Amazon reduces purchases for its data center infrastructure.
    Chain: Amazon redesigns data center hardware around custom silicon → Power and analog content designed in internally → Merchant analog sockets per server decline → Texas Instruments data center revenue slips
    What would break the call: Denser accelerator racks raise power stage content per board, lifting TI analog dollar content instead.
  18. ▲ KEYS Keysight Technologies — impact +22 · confidence low · horizon quarters · 3rd order
    Custom silicon bring-up and high speed interconnect validation increase demand for Keysight test and measurement instruments.
    Chain: Hyperscalers expand internal chip design programs → More silicon bring-up and compliance testing needed → Keysight high speed digital instrument orders rise → Electronic industrial solutions revenue grows
    What would break the call: Capital equipment budgets tighten and hyperscalers reuse existing test benches instead of buying new instruments.
  19. ▼ SMCI Super Micro Computer — impact -20 · confidence low · horizon quarters · 3rd order
    Supermicro sells fewer merchant GPU systems as Amazon builds accelerator racks around internally designed chips.
    Chain: Amazon builds racks around internally designed accelerators → ODM direct manufacturing handles AWS rack assembly → Merchant GPU system demand at Supermicro softens → Server revenue growth and margin compress
    What would break the call: Supermicro wins large neocloud or sovereign AI orders that offset any hyperscaler merchant system decline.
  20. ▲ GOOGL Alphabet — impact +18 · confidence low · horizon quarters · 3rd order
    Amazon validating in-house accelerators reinforces the TPU strategy that lowers Google Cloud compute cost per workload.
    Chain: Amazon deepens internal silicon commitment → Market revalues hyperscaler custom accelerator strategies → Google TPU roadmap gains comparative credibility → Cloud margin narrative strengthens
    What would break the call: Customers reject non-CUDA accelerators at scale, forcing both Google and Amazon back toward merchant GPUs.
  21. ▼ AVGO Broadcom — impact -15 · confidence low · horizon quarters · 2nd order
    Broadcom competes for the same hyperscaler custom silicon dollars, and stronger AWS internal capability limits its addressable sockets.
    Chain: Amazon builds stronger internal design capability → Need for Broadcom custom XPU partnership diminishes → Broadcom addressable hyperscaler ASIC sockets narrow → AI semiconductor revenue pipeline tightens
    What would break the call: Broadcom discloses AWS as a new multibillion dollar custom XPU customer for Trainium generations.

Sources

What happened

HorizonVerdictTop pickExcess vs SPY
1 trading daysRightSNPS+12.6% vs SPY
5 trading daysRightSNPS+13.7% vs SPY

More

Related scenarios

Stocks in this scenario

SNPS · AMZN · TSM · ARM · INTC · NVDA · CDNS · AMD · ALAB · MU · AMKR · CRDO · MRVL · ALGM · ANET · CRWV · TXN · KEYS · SMCI · GOOGL · AVGO

News events · All scenarios · Pricing · Track record · Methodology · Map your own scenario