Tech Brew Ride Home

Episode: Nvidia Kindacquires Groq
Date: December 29, 2025
Host: Brian McCullough

Overview

This episode dives into Nvidia’s unusual “kind-of-acquisition” of AI chip startup Groq, the strategic rationale and implications for the chip industry, and what this move says about the fiercely competitive landscape in AI hardware. The episode also touches on Samsung and Google’s partnership to bring Google Photos to TVs, the end of remote exams in accounting due to AI cheating, and the actual (somewhat underwhelming) state of robotics ahead of CES 2026.

Key Topics & Insights

1. Nvidia’s “Acquisition” of Groq

[00:35–11:55]

• Deal Structure and Terms:

  • Nvidia and Groq announce a licensing deal, with Nvidia paying for the right to use Groq’s tech.
  • Notably, CEO Jonathan Ross (ex-Google TPU architect) and other executives (around 90% of staff) will join Nvidia.
  • Groq remains independent, with a new CEO, but transfers key people and IP access.
  • “Another big acquisition in all but name. Nvidia has agreed to a licensing deal with Groq… around 90% of Groq’s employees will be joining Nvidia. Groq nonetheless says it will continue operating independently.” – Brian [00:40]

• Why Nvidia Did This:

  • Groq’s “low latency chips are extremely responsive” and can bolster Nvidia’s offering.
  • Strategic talent grab: Ross is the mind behind Google’s Tensor Processing Unit (TPU), and Nvidia is clearly positioning to counter Google’s momentum in specialized AI chips.
  • Regulatory angle: By structuring this as a licensing deal (not a classic acquisition), Nvidia possibly sidesteps antitrust scrutiny—important, as Nvidia controls over 90% of the AI chip market.

• Details on Groq & Its Tech:

  • Founded in 2016 by Jonathan Ross, who was key in developing Google’s first TPU.
  • Groq focuses on “Language Processing Units” and has built a cloud business for smaller developers.
  • Their architecture keeps more data on chip using SRAM (Static Random Access Memory) instead of high-bandwidth external memory, potentially making them faster for certain types of AI workloads.
  • “Groq’s low latency chips are extremely responsive to inputs and will add new capabilities to Nvidia’s products and open up new areas of the market, Nvidia said.” – Brian [02:49]

• Financial Details:

  • Sources: Groq’s execs, employees, and investors (including BlackRock and Tiger Global) get payouts as if at a $20B valuation—a massive jump from Groq’s last public $6.9B valuation.
  • “Most shareholders will apparently be getting per share payouts at a $20 billion valuation, while Groq employees will be paid cash for all of their vested shares.” – Brian [04:23]

• Market and Industry Context:

  • As Big Tech giants (Google, Microsoft, Amazon) push in-house for AI chips, Nvidia secures both technology and talent to maintain its dominance.
  • The deal is compared to how Meta handled Scale AI: strategic investment, licensing IP, and talent acquisition.

2. Industry Analysis & Reactions

[07:02–11:06]

• Gavin Baker on X (Twitter):
  Details the technical rationale—SRAM-based (on-chip memory) architectures like Groq’s excel at inference/low-latency workloads.

  • “SRAM architectures can hit token per second metrics much higher than GPUs… Extremely low latency per individual user at the expense of throughput per dollar.”
  • “It is now abundantly clear from Cerebras and Groq’s recent results that users are willing to pay for speed…” – Gavin Baker, paraphrased by Brian [08:08]

• MG Siegler’s Commentary:

  • Suggests the main goal is acquiring Jonathan Ross and Sonny Madra (Grok president), not just the IP—calling attention to the talent war in AI chips.
  • Notes the “non-exclusive licensing” is a regulatory fig leaf: “But is anyone else really getting access to this IP now?”
  • “Nvidia probably feels confident that with Ross and Madra, not to mention their own in house prowess, they’ll be able to implement it and execute upon it far better than anyone else, and they’re undoubtedly not wrong.” – MG Siegler, read by Brian [09:50]
  • Points out the deal’s $20B price tag with “Nvidia is paying $20 billion to grab some talent and license some tech. $20 billion. It’s one of the largest deals of any sort in the history of deals, and they’re technically acquiring nothing.” [10:40]
  • Speculates if this deal pushes regulators to step in and notes the urgency to complete such a move before a potentially less tech-friendly administration.

3. Other Headlines

a. Samsung x Google: Google Photos on Tizen TVs

[16:45–18:55]

• Samsung will integrate Google Photos into its Tizen OS TVs, launching in March 2026—exclusive for the first six months.
• Curated “memories” will be shown, not the full Google Photos library at launch.
• Later, “Create with AI” tools will roll out, bringing themed templates and personalized experiences.
• “Given Google Photos’ popularity, its integration into Samsung TVs makes a lot of sense. It brings one of Google’s most used services to the biggest screen in the house…” – Brian [18:50]

b. Accountants End Remote Exams Due to AI Cheating

[19:00–21:14]

• The Association of Chartered Certified Accountants (ACCA) will end remote exams due to rising AI-enabled cheating.
• Exams will revert to in-person starting March.
• “We’re seeing the sophistication of cheating systems outpacing what can be put in in terms of safeguards,” says ACCA CEO Helen Brand. [20:44]

c. Robotics: The Hype vs. Reality Ahead of CES

[21:15–End]

• Viral video: Man “testing” a humanoid robot, gets kicked—highlighting robotics aren’t quite assistant-ready.
• Robotic execs emphasize the gap between current demos and practical utility—most humanoid bots are best for simple, repetitive tasks, not generalized household help.
• “We’re doing a big extrapolation from watching videos of robots doing laundry to a butler in my house that can do everything.” – Annie Kelkar, McKinsey [22:31]
• $5B invested in humanoid robots this year, but genuine “robot butlers” remain far off.
• Battery and motor advances are noteworthy, but commercial viability is narrow for now.
• “This will be the year of robotics at CES, the first real domestic robots, even if they’ll still be early and expensive. So let’s see if that actually pans out.” – Brian [24:31]

Memorable Quotes

• On Nvidia’s Motivation:
  “Maybe Nvidia is fully bought in on the former. Or maybe they simply want to access the key talent that created the TPU and importantly to keep them away from anyone else who might try to make their own XPUs.” – Brian, paraphrasing MG Siegler [11:08]

• On Market Dynamics:
  “In an era of growing energy fears, this is potentially a real problem for Nvidia… They probably need to have their own TPU-like option, even if they remain fully committed to their GPUs as being the bigger and better option. Certainly they need to counter this narrative on the inference side.” – Brian [11:38]

• On Technology Hype:
  “We’re doing a big extrapolation from watching videos of robots doing laundry to a butler in my house that can do everything.” – Annie Kelkar, McKinsey [22:31]

Timeline of Major Segments

• 00:35 — Nvidia "kind-of-acquires" Groq: Deal details, company backgrounds
• 04:23 — Financials, licensing and payouts; comparison to other tech M&A/partnerships
• 07:02 — Gavin Baker’s commentary: Why low-latency matters in AI inference
• 09:50 — MG Siegler’s perspective: It’s about the talent; possible regulatory challenge
• 16:45 — Samsung & Google Photos partnership for TVs
• 19:00 — ACCA bans remote exams due to AI cheating
• 21:15 — The robotic realities: Limitations of today’s humanoid bots, CES 2026 preview

Tone

The episode features Brian McCullough’s chatty, savvy tech-insider delivery. He weaves in expert quotes, provides context and speculation, and isn’t afraid to poke fun at both hype cycles and regulatory gamesmanship—giving listeners a brisk but thoughtful rundown of the tech stories that matter.

For those who missed this episode: it's a clear, engaging look at a game-changing deal in AI hardware, how Nvidia is playing chess with its rivals, the ongoing spread (and, sometimes, retreat) of AI in the real world, and the ever-present gulf between what robotics videos promise and actual product reality.