The Byrne-Wheeler Report

In this episode, we explore the latest developments in AI hardware, including NVIDIA's NV HBM, Astera's Leo family, Velaura's pivot to robotics and low-power design, and insights from MLPerf benchmarks. If you're a semiconductor or AI professional, join us for a discussion of the future of AI processing, communications, and industry trends. 00:00 Introduction and episode overview

01:06 NVIDIA's NV-HBM and custom memory interface

03:16 Moving memory controllers to base die and implications

06:31 Astera's second-generation CXL memory controllers

09:17 Are CPUs made for agentic AI just hype?

13:55 Is local AI worthwhile?

15:06 What is the CPU market TAM?

19:17 Having received a Series A, what is Velaura doing?

25:19 MLPerf benchmarks and AMD's performance

31:25 What's the Qualcomm-Huawei agreement about?

32:55 What's the Qualcomm-Amazon agreement about?

38:48 Arm's new CPU cores and future prospects

41:27 Closing remarks and upcoming events

Music licensed through Soundstripe.

Code: GHYB3TSCQYEN8ICL

Caliente and Picante! Joe and Bob Inside Hot Chips 2026 to Find The Designs Shaping AI Semiconductors.

Join Byrne-Wheeler Report hosts Joe and Bob as they discuss key disclosures from Hot Chips 2026, including Intel’s Diamond Rapids Xeon, AMD’s Helios MI450 platform, NVIDIA’s LPX/LPU architecture, Cerebras’s CS-4 wafer-scale engine, and OpenAI’s Jalapeño ASIC. If you work in AI hardware, semiconductors, GPUs, CPUs, interconnects, or data center architecture, this episode delivers insight unavailable in press releases vendor pitches.

00:00 Episode Highlights

00:53 Introduction and overview of Hot Chips 2026

02:51 Intel Crescent Island GPU

08:05 Intel Diamond Rapids CPU

13:59 AMD Helios and MI400

18:22 Nvidia Scale-In Networks, Bluefield 4 DPU

21:10 Nvidia Groq LPX LP30

29:20 Cerebras WSE-4

34:23 OpenAI Jalapeno

39:23 Goodbyes

https://wheelersnetwork.com

https://XPU.pub

Music licensed through Soundstripe.

Code: T1SPLXVECWXDHVM1

In Episode 16 of the Byrne-Wheeler Report, Joe and Bob discuss the latest developments in AI hardware, including disclosures at AMD's Advancing AI day, NVIDIA Vera, and upcoming conferences Hot Chips and OCP APAC. They analyze new processor architectures, interconnect standards, and market implications for AI and data center computing.

00:00 Introduction and episode overview

00:24 Upcoming Conferences: Hot Chips and OCP APAC

01:07 Intel Processor and GPU Disclosures at Hot Chips

04:24 AMD AI Day: Helios Wide Rack and Interconnect

06:54 AMD AI Day: EPYC Venice Gen 6 Processors

10:37 AMD AI Day: Verano and XGMI/UALink/SUE-T Interconnect for Helios 500

12:09 AMD AI Day: GPU Roadmap--MI 500 Scales Beyond 72 GPUs and Ratchets Up Performance

15:31 AMD AI Day: MI455X GPU Architecture Improvements and Performance vs Nvidia Rubin

16:44 AMD AI Day: Pensando Roadmap and Helios Scale-Out Bandwidth

18:03 AMD AI Day: Microsoft Azure and Anthropic Customer Announcements

20:07 Alphabet/Google Earnings and AI Infrastructure Investments

22:37 Introduction to Astera Labs Taurus Re-Timers and Re-Drivers

26:07 NVIDIA Vera: A High-Bandwidth CPU for GPU Hosts and App Sandboxing

30:13 Closing Remarks and OCP-APAC, Hot Chips Reminders

Joe: https://xpu.pub

Bob: https://wheelersnetwork.com

#AI hardware, #AMD, #NVIDIA, Hot Chips, #OCP APAC, processor architecture, interconnect standards, data center, AI inference, #GPU

AMD's AI Day 2026 - https://www.amd.com/en/corporate/events/advancing-ai.html

NVIDIA Vera Platform - https://www.nvidia.com/en-us/data-center/vera-cpu/

Hot Chips Conference - https://www.hotchips.org

OCP APAC Summit - https://www.opencompute.org/summit/2026-ocp-apac-summit/theme

Astera Labs Taurus Re-timers and Re-drivers - https://www.asteralabs.com/news/astera-labs-expands-taurus-family-with-industrys-first-ocp-standard-footprint-compatible-3-2t-smart-retimers-and-smart-redrivers/

Google Cloud Earnings Report Q2 2026 - https://abc.xyz/investor/earnings/

Music licensed through Soundstripe.

Code: J5RRCCF40OGJOX7B

Byrne and Wheeler discuss recent developments in the tech industry, including #AMD's acquisition of #MEXT, #NVIDIA's impressive growth, #Broadcom's earnings, and innovative chip technologies like #Tensordyne’s log-domain processing and #Huawei’s Tau scaling. They analyze market trends, product launches, and strategic partnerships shaping the future of semiconductors and AI hardware.

00:00 Introduction and MEXT Acquisition

02:16 Earnings Reports: NVIDIA's Growth and Strategy

06:14 Broadcom's Results and Market Dynamics

08:14 #Broadcom Tomahawk, #Marvell Teralynx, Cisco One Switches and Next-Gen Serdes

14:17 Tensordyne

15:46 Huawei Tau Scaling and Logic Folding

20:18 Nvidia at GTC and with Marvell at #Computex

24:40 #Intel and #SambaNova Announce Collaboration at Computex

Joe and Bob discuss the latest advancements in semiconductor technology, including IBM's sub-1 nanometer process, Intel's new milestones, Qualcomm's data center strategy, Etched, OpenAI, and Oxmiq, interconnect companies/technologies such as CXL and Upscale, and memory company Micron Technology.. They explore how these innovations are shaping the future of chip design and data center infrastructure.

00:00 Introduction and Weather Banter

01:11 IBM's Sub-1 Nanometer Technology Announcement

05:37 Intel's 18A-P Process Updates

08:58 Qualcomm's Investor Day Highlights

13:35 Modular AI Software Stack Discussion

15:42 Qualcomm's C1000 CPU Roadmap

19:36 Qualcomm ASIC Business Growth

22:07 MediaTek's AI Data Center Revenue

24:07 Emerging Startups in the XPU Space: Etched

26:19 OpenAI/Broadcom's Jalapeno ASIC Announcement

29:25 Upscale Secures Funding

30:16 Micron's Earnings and Business Model

31:58 Oxmiq Labs and GPU Innovations

32:59 CXL Technology Resurgence: Meta and Panmnesia

This episode of The Byrne-Wheeler Report explores topics in computer memory, particularly with respect to AI processing and servers. Hosts Joe and Bob begin with a discussion of Caltech spinout PrismML’s 1-bit weights and Google's TurboQuant method of reducing KV cache entries to 4 bits, discussing their implications for on-device AI and memory efficiency.

Next, the hosts welcome Gary Smerdon, CEO and founder of MEXT to discuss the company’s innovative software solution to reducing DRAM costs and improve memory utilization in servers. Tune in to learn how AI-driven memory management can transform data center efficiency and lower expenses.

Finally, Jim Handy, a renowned memory analyst, joins Joe and Bob to discuss the recent dramatic rise in memory prices, its impact on tech products, and the future outlook of the memory market amid AI-driven demand and manufacturing challenges.

The conversation covers a range of topics.

• Amazon's internal chip business is making a significant impact on the merchant market
• AMD's data center business has exceeded that of Intel's for the first time
• Google's TPU V8 introduces a new topology for inference clusters
• Astera Labs's Scorpio X switch targets AI scale-up fabric
• RISC-V companies are diversifying into chip business and adding silicon to their portfolios, indicating IP-business challenges, particularly in Europe

Chapters

• 00:00 Introduction and News Items
• 07:19 AMD's Earnings Call
• 13:32 Google's TPU V8
• 20:37 Astera Labs's Scorpio X Switch
• 26:36 RISC-V Related Announcements

So much news from GTC and OFC! Covered companies include AMD, Tesla (Terafab), Cerebras, Amazon, AWS, Upscale AI, Nvidia, Eridu, Meta, Broadcom, and Arm.