Recorded June 15, 2026, Episode 14 covers: NVDA, AVGO, MRVL earnings Marvell T100 102.4Tbps Ethernet switch Tensordyne logarithmic math Huawei Tau scaling (roadmap to 14A-equivalent density) GTC Taipei RTX Spark announcement Marvell keynote at Computex with Jensen cameo Intel+SambaNova AI inference racks plus Clearwater Forest Congratulations to MEXT on being acquired by AMD. Good things come to our sponsors 😀

This episode features an in-depth discussion on the latest developments in data-center chips, AI accelerators, interconnect technologies, and RISC-V ecosystem growth. Joe and I analyze AMD and Intel earnings, Google TPUs, Astera Labs, and emerging startups. (Recorded May 6.)

As I expressed at the end of this episode, memory is a world unto itself. A big thanks to Jim Handy for helping us understand what's going on with DRAM, and to MEXT for being our first sponsor. Bringing an AI engine into memory tiering is a major leap over existing techniques like TPP. Even better is the fact that MEXT's approach requires no new hardware, unlike CXL-based memory expansion. Jim Handy explains why HBM consumes 3x the wafers of equivalent DDR bits, which is known as the trade ratio. Joe and I also discuss TurboQuant and what's happening with AI quantization. An important distinction is that TurboQuant addresses the KV cache, not model weights. Its first application will be for on-device inference, meaning it doesn't take any pressure off HBM demand in the near term.

In this episode, Joe and I tried a little different format, doing some quick-hit news items in the first 5 minutes before diving into items that warranted more analysis. My networking audience may want to skip to 4:21, where we discuss the Nvidia and Upscale AI announcement around Spectrum-X. Next, we discuss startup Eridu's funding and product positioning. From OFC, I cover the new Optical Compute Interconnect Multi-Source Agreement (OCI-MSA) and Broadcom's surprise announcement of Taurus, the industry's first 400G/lane PAM4 optical DSP. From GTC, we discuss the addition of Groq plus new disclosures around Rubin Ultra, Feynman, and the Kyber rack.

Joe and I did another speed round as we tried to catch up with all of the news in the AI chip world. For my networking audience, you can skip to 22:07 to catch Ayar Labs' funding, followed by Nvidia and Broadcom earnings commentary. If you're interested in all the recent hyperscaler deals, you can start at the beginning. We cover Nvidia's Grace at Meta, the 6GW AMD-Meta deal, and OpenAI deals with Amazon and Nvidia. Chip-startup funding rounds include MatX, Talaas, HyperAccel, and SambaNova, as well as Ayar. If you missed Episode 9, you can find a playlist here:  https://youtube.com/playlist?list=PLg8_XIsi4WPd5CAOGgLTfGO7lpXKNjtvf&si=8tQloJuNsIpVs37j 

The latest episode of The Byrne-Wheeler Report covers a slew of deals from funding to outright acquisitions. In my coverage space, Upscale AI has raised another $200 million at a unicorn valuation. Marvell is acquiring CXL-switch startup Xconn Technologies. Joe and I discuss two announcements in the RISC-V arena. In case you're wondering about Episode 7, it was a special episode with guest Mike Demler discussing Nvidia's Christmas Eve $20 billion bombshell Groq deal.

We recorded Episode 6 of The Byrne-Wheeler Report on December 19, 2025, i.e., before Nvidia's Christmas Eve bombshell (Groq). More on the latter to follow soon. Meanwhile, Joe and I caught up the latest on Google TPU customers, Tsavorite coming out of stealth, Marvell's Industry Analyst Day disclosures, and Qualcomm acquiring Ventana.