Semi Doped

Why Marvell Wants Lumentum: Vertical Integration in AI Optics With Cignal AI

How Intel Took EMIB From Science Fiction to High Volume: Intel Foundry's Ravi Mahajan

EMIB (Embedded Multi-Die Interconnect Bridge) technology provides a scalable, high-density solution for heterogeneous chip integration by utilizing localized silicon bridges within organic substrates. This approach decouples high-speed I/O from power delivery, allowing for fine-pitch interconnects only where necessary ...

RF Over Plastic Is a Third Option Between Copper and Optics: Point2's Sean Park

MicroVCSELs vs MicroLEDs, a MicroExplainer

Short-reach data center interconnects are transitioning from microLEDs to microVCSELs to address the limitations of copper and the scaling requirements of AI systems. While microLEDs rely on spontaneous emission and suffer from chromatic dispersion that restricts their reach to approximately five meters, microVCSELs ut...

AI Inference Is a Data Movement Problem: Microsoft's Saurabh Dighe on Maia 300

Microsoft’s Maia 300 architecture represents a strategic shift toward inference-first AI infrastructure, designed to maximize token throughput and energy efficiency for planet-scale workloads. Saurabh Dighe, Corporate Vice President of Azure Hardware Systems, details how the system leverages a software-defined data flo...

Synopsys’s Amazon Deal and GPT-Synopsys: Ravi Subramanian

Synopsys is transforming semiconductor design by shifting toward application-optimized IP and integrating agentic AI into the EDA workflow. The recent $1 billion-plus deal with Amazon highlights this transition, where custom, workload-specific IP replaces traditional off-the-shelf standards to meet the extreme performa...

Every Muse User Gets 2 CPUs and 8GB in Meta's Cloud. Your Phone Already Has That

Flash Hit Its Physical Limit. ReRAM Is the Replacement: WeeBit Nano's Ilan Sever

Qualcomm's Durga Malladi: HBC vs HBM, Dragonfly AI 250, and Winning on TCO

High Bandwidth Compute (HBC) serves as a critical technical differentiator for Qualcomm’s data center strategy, addressing the growing "memory wall" where compute capabilities outpace memory bandwidth. By utilizing near-memory computing, HBC performs processing directly adjacent to stacked DRAM, significantly reducing ...

Pacing AI Means More Compute, Not Less

TSMC Will Buy High NA After All: ASML's New 6x12-Inch Mask

OpenAI’s Jalapeño! Feeling Hot Hot Hot!

Grok Bots and How CPUs are used in Agentic AI

The emergence of agentic AI platforms like Grok bot signals a shift toward cloud-based virtual machine environments, simplifying access for users while significantly increasing hardware demand. This transition redefines the relationship between compute components, where GPUs function as the "genius" performing high-lev...

Tensordyne's R K Anand: HPE Juniper Fabric, Logarithmic Math, MoE Inference, Air Cooling, 3nm

NEWS TAKE: China's Optical Ban, Volta's $10B Anthropic Deal, AMD's Earnings

GlobalFoundries Thomas Barber: CPO, Silicon Photonics, 300mm, SiGe, OCI, NRZ

How Retimers Built an $80B Company: The Story of Astera Labs

Astera Labs has transformed from an obscure startup into a critical player in AI infrastructure by mastering the challenge of high-speed signal conditioning. As data transfer speeds for PCIe Gen 5 and beyond push copper traces to their physical limits, re-timer chips are essential for maintaining signal integrity by mi...

3 Stories Behind the Semi Selloff: Hyperscaler CDS, SK Hynix Earnings, China's DUV

Datacenter Interconnects: Copper vs. Optics, Nvidia's 78-Layer PCB, Co-Packaged Optics (CPO)

PicoJool's Al Yuen: The Case for GaAs VCSELs in Scale-Up Interconnects