Amazon has entered into a $1 billion multi-year intellectual property licensing agreement with Synopsys to support the development of custom semiconductor designs. The arrangement establishes Amazon as a flagship custome…
#Custom Silicon
This week in Custom Silicon · updated Thu Oct 08 2026
Major cloud providers are deepening investments in proprietary semiconductor development this week. Amazon committed $1 billion to Synopsys for silicon IP licensing to accelerate custom chip design for data centers, while Semifive secured a $52 million accelerator contract with a US hyperscaler for 2028 production. These deals reflect intensifying competition among cloud giants to differentiate hardware for specialized workloads. Separately, quantum computing research advanced with new error-correction techniques combining surface codes and IBM's architecture, though decoding remains computationally challenging as systems scale. These concurrent developments underscore how infrastructure providers are pursuing custom silicon across both classical and quantum domains.
AI-written weekly briefing drawn from this topic's recent stories.
Compute · Open in RealityHacker · RSS
Amazon commits $1 billion to Synopsys for silicon IP licensing in multi-year deal
Custom silicon designer Semifive lands major accelerator contract from undisclosed US hyperscaler
Chipmaker Semifive has secured a $52 million contract with an undisclosed American AI company to design specialized accelerators for deployment at scale. The chips are slated to enter production in 2028 and will target h…
Zotac RTX 3060 Warranty Dispute: Import Date Takes Precedence Over Purchase Date in India
A Zotac RTX 3060 owner in India purchased their graphics card in June 2025 but was denied warranty coverage after one year despite Zotac's advertised three-year warranty policy for the region. The manufacturer claimed th…
Surface Code Meets IBM Heavy-Hex in New Quantum Error Correction Research
Quantum Elements and University of Southern California researchers used the Orbit error-suppression tool in work that combines surface code with IBM's heavy-hex architecture. Published in Nature, the research aims to imp…
IonQ Pitches CPU-Centric Decoding for Quantum Error Correction
Quantum computing still faces a major obstacle in decoding error syndromes, since qubits cannot be measured directly without losing their state. The computational load of decoding grows with system size and depends on th…
Brazilian institute to host IQM's Spark quantum machine
IQM, a Finnish quantum computing company, will install its Spark system at the Eldorado Research Institute in Brazil. This marks the first deployment of a quantum computer by the firm in South America.
Mitsubishi backs optical quantum computing firm OptQC
Mitsubishi has made an investment in OptQC, a Japanese startup developing optical quantum computing technology. The financial terms were not disclosed. The move signals growing interest in quantum computing alternatives.
Broadcom Forecasts 88-Fold Surge in Custom AI Chip Revenue by 2028
Broadcom projects its custom AI accelerator and networking chip revenue will leap from $2.6 billion in fiscal 2022 to $230 billion by fiscal 2028, a two-order-of-magnitude increase. The company has secured HBM memory, wa…
Z-Order Memory Addressing Proposed to Overcome the Memory Wall in High-Performance Computing
A new approach to memory addressing using Z-Order (Morton) layout could eliminate the memory wall that limits modern server processors. The method aims to reduce cache coherency traffic and keep cores busy by aligning me…
Infleqtion and Nvidia Shrink Qubit Overhead for Quantum Error Correction
Infleqtion and Nvidia have combined their quantum software and orchestration tools to lower the number of physical qubits required to form a logical qubit, a critical step for error correction in quantum systems. The int…
DRAM designer Piecemakers targets edge AI with hybrid-bonded memory instead of HBM
PieceMakers, a Nanya-backed DRAM designer, began trading on Taiwan's Emerging Stock Board on September 16, betting that AI inference memory will diverge from HBM. The company proposes stacking DRAM directly on the proces…