What's Different About HBM4
Memory bandwidth is limiting the flow of huge datasets that are needed to train AI models. There is much more data to process, store, and retrieve, but the speed at which that data moves through high-bandwidth memory (HBM) stacks is significantly lower than the speed at which data can be processed. Frank Ferro, group director for product management at Cadence, talks with Semiconductor Engineering about the new HBM4 standard, how it compares with HBM3, and what will be needed to meet the insatiable demands of hyperscaler data centers in the future.
Related Semiconductor IP
- HBM4E PHY and controller
- HBM4 Memory Controller
- HBM4 PHY IP
- HBM4 Controller IP
- TSMC CLN3FFP HBM4 PHY
Related Videos
- Ask the Expert: HBM4 Controller IP
- Weebit Nano is scaling up for 2026: The Year of the Tornado
- Weebit Nano on why the next NVM is ReRAM
- Intel Foundry Direct Connect: Mahesh Tirupattur Talks About Power
Latest Videos
- Ask the Experts: Caliptra and Root of Trust for AI Infrastructure
- Cadence UCIe at 64 GT/s in Silicon: Why It Matters for Your Next Chiplet Design
- Can AI Be Trusted? Synopsys on Agentic AI and Autonomous Engineering
- Caliptra Silicon Root of Trust in BMC SoCs
- UALink 2.0 Specification Update: Advancing Accelerator Interconnect Technology for AI Workloads