Jul 5, 2026 · 1h 11m · david-senra
Why Asking the Right Questions Is the Most Important Skill in the AI Age
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
David Senra interviews Groq founder Jonathan Ross on the architectural breakthroughs of Language Processing Units (LPUs), the transformative impact of ultra-fast inference speed on artificial intelligence, and the leadership principles, talent frameworks, and cognitive skills essential for navigating the AI era.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. David holds 21.5% of the talking time here. How this is scored →
David as informed peer 4.9 Guest teaching 5.3 Guest disagreement 0.4 David pushing back 0.3
speaking balance: gold is David, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 22:53 Ross mocks venture capital lemming behavior
Ross bluntly dismisses conventional West Coast venture capital psychology, mocking them as lemmings who blindly follow or reject deals without independent analysis.
Hardest push from David ▶ 42:00 Senra interrupts to demand why fast inference was dismissedSenra halts the narrative flow to challenge Ross on why everyone in the industry supposedly failed to see the obvious value of fast inference just four years earlier.
Biggest teaching moment ▶ 31:00 Ross explains how hardware speed creates deeper AI reasoningRoss delivers a technical masterclass using DeepMind's AlphaGo match, explaining how TPU compute velocity directly uncovered Move 37 by allowing deeper tree search.
David holds their own ▶ 1:01:55 Senra masterfully connects manufactured discontent across four historical giantsSenra showcases formidable historical mastery by seamlessly connecting Ross's concept to Tim Grover on Michael Jordan, David Ogilvy, Steve Jobs, and Edwin Land.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | David as informed peer | Guest teaching | Guest disagreement | David pushing back | Why |
|---|---|---|---|---|---|---|
| The NVIDIA Partnership and GPU-LPU Architecture Synergy | 4 | 5 | 1 | 0 | Senra opens by asking about the reported Nvidia partnership and references a previous private discussion at Nvidia GTC regarding complementary architectures. Ross explains the technical synergy between GPUs and LPUs across compute vs memory throughput bottlenecks and details agentic AI workloads. | |
| Personal Prototyping and Dynamic AI Information Briefings | 5 | 3 | 1 | 0 | Ross details his personal hobby projects and his custom AI daily briefing. Senra demonstrates expertise by comparing Ross's system directly to Spotify co-CEO Gustaf's curated AI podcast workflow, prompting Ross to explain why interactive questioning beats static audio. | |
| Asking the Right Questions in the AI Era | 3 | 4 | 0 | 0 | Senra quotes one of Ross's tweets about the transition from answering questions to asking questions in the AI age before running a sponsor ad. Ross elaborates on how individual contributors are transforming into leaders who direct AI through targeted questioning. | |
| First Principles of Leadership and Extreme Delegation | 6 | 4 | 1 | 1 | Ross outlines his extreme delegation leadership style and non-control freak nature. Senra enriches the dialogue by quoting Shopify's Tobi Lütke and UFC's Dana White on self-awareness, validating Ross's tailored approach to organization design. | |
| Management Maturation and Under-Constraining Goals | 6 | 5 | 0 | 0 | Ross reflects candidly on early management failures at Grok and the importance of giving teams under-constrained, clear goals. Senra reinforces the concept by quoting Kelly Johnson from Lockheed Skunk Works on extreme performance springing from one brutally clear priority. | |
| Jensen Huang's Operating System and Founder Confidence | 7 | 5 | 0 | 0 | Ross explains how Jensen Huang runs Nvidia without one-on-ones to eliminate internal politics and shares how shadowing a seasoned executive built his leadership confidence. Senra references his comprehensive study of Huang across multiple podcast episodes and draws comparisons to Jeff Bezos. | |
| Managing Deep Talent and Venture Capital Psychology | 4 | 6 | 2 | 1 | Ross breaks down the difficulty of managing research scientists and contrasts herd mentality among West Coast VCs with independent East Coast crossover investors using the Keynesian beauty contest analogy. Senra presses on Grok's past financial peril and notes the irony of Silicon Valley funds missing the Nvidia transaction. | |
| The Breakthrough Hybrid LPU-GPU Innovation | 4 | 6 | 0 | 1 | Senra asks for the backstory of the hybrid chip architecture breakthrough. Ross explains how Grok's COO initiated combining GPUs and LPUs for token prefill versus token generation, which led directly to the rapid Nvidia licensing deal. | |
| Inference Speed as Intelligence and the AlphaGo Breakthrough | 3 | 7 | 0 | 0 | Ross delivers an in-depth lesson on how TPU hardware speed enabled AlphaGo's jump in Elo rating and the discovery of Move 37 against Lee Sedol. He illustrates why fast inference fundamentally enhances reasoning quality rather than just delivering rapid text output. | |
| Sponsor Message: AppLovin Advertising Platform | 3 | 6 | 0 | 0 | Following a sponsor read, Senra asks about Ross's concept of Reality Quotient (RQ). Ross defines RQ as the capacity to identify and play the dominant game in an industry and details how founders must practice change management by framing major shifts as continuity. | |
| Return on Luck and the Validation of Fast Inference | 5 | 6 | 1 | 1 | Ross recounts how Grok repeatedly missed LLM tailwinds before committing to ultra-fast inference and going viral. Senra pushes Ross to explain why the entire AI industry failed to appreciate inference speed four years ago. | |
| Intentional Leadership and Overcoming Team Friction | 6 | 5 | 0 | 1 | Ross explains adopting David Marquet's intentional leadership model ('I intend to') to overcome internal organizational pessimism. Senra demonstrates subject familiarity by referencing the underlying submarine case study and clarifies whether Ross faced pushback from teammates or clients. | |
| Sponsor Message: Deel Global Hiring Platform | 6 | 5 | 0 | 0 | Senra connects Ross's Grok Bonds initiative to Phil Knight's employee debt conversions in Shoe Dog. Ross details how voluntarily cutting salaries down to statutory minimums preserved compiler development and kept the company afloat. | |
| The People Spec and Hiring for Loss Bias | 5 | 6 | 0 | 0 | Ross explains the People Spec, shifting hiring criteria from screening for positive traits to filtering out fatal negatives, and hiring for loss bias. Senra validates the framework by quoting Grok's core design philosophy. | |
| Manufactured Discontent and the Race for Global Compute | 8 | 5 | 0 | 0 | Ross describes manufactured discontent and his personal sense of urgency regarding global compute shortages. Senra provides an impressive sequence of historical founder parallels, referencing Tim Grover's study of Michael Jordan, David Ogilvy's divine discontent, Steve Jobs' creative drive, and Edwin Land's headlight safety metrics. | |
| The Near-Zero Cost of Code and the Future of Education | 4 | 6 | 0 | 0 | Senra invites an optimistic outlook on AI to counter public backlash. Ross predicts the collapse of marginal coding costs, compares software creation democratization to historical literacy expansion, and outlines how education should be redesigned around inquiry rather than memorization. |
Statements from this episode (28)
Ross: Groq's NVIDIA deal closed within three weeks of first call
“The call where the idea was first floated was about three weeks before money was in the bank.”
Ross: Combining GPUs and LPUs yields superior performance across curves
“And so GPUs and LPUs combined ended up giving better performance across the performance curves.”
Ross: Groq initially approached Jensen Huang to buy 100,000 GPUs
“We had implemented it and we had gone to Jensen asking, could we buy about a 100,000 GPUs because we were going to deploy them ourselves.”
Ross: Agentic AI drives exponential growth by delegating tasks to other AIs
“AI is really good at using AI. And that's really what agentic is, right? So humans benefit from using AI. So does AI. Just like you would do research about me before I show up on your show, AI is going to kick off a job doing research on different tools it's g…”
Ross: Machine micropayments will cause autonomous payment volumes to skyrocket
“Yeah, I think it's still early, and one of the limiters is payments aren't really built for this yet, but if you can make micropayments, the number of payments is going to skyrocket.”
Senra: Spotify Co-CEO Uses AI Agents to Generate a Private Daily Podcast
“I just spent time and recorded an episode with Gustaf, who's the co-CEO of Spotify right now. And he did something very similar because he tries to avoid any kind of feeds. The whole thesis behind, like, the organizing principle of Spotify is, like, time well …”
Ross: Workers Are Shifting from Individual Contributors to Leaders of AI
“Yeah, and this also goes towards a shift of going, people are moving from being ICs or individual contributors to all being leaders, but leaders of AI.”
Ross: Groq hires highly autonomous people who would fail in corporate setups
“When I hire people, I hire very autonomous people who often go off and execute on their own, and they would be terrible in most corporate environments.”
Ross: Learning to manage people cost Groq three to four years
“The first thing that you have to do as a founder is you have to go from the technical thing that you know how to do and that you can add value with to learning how to manage people. And for me, that probably cost Grok three to four years.”
Ross: Groq gave every employee a challenge coin targeting 25M tokens/sec
“So everyone at Grok had a challenge coin that said, twenty-five million tokens per second and had a little graph of it going up, and everyone knew that was the thing to do.”
Ross: Jensen Huang never holds private one-on-one meetings at NVIDIA
“There is no circumstance at NVIDIA where there, where Jensen has one-on-ones with people and tells them one thing.”
Ross: Replace one-on-ones and private emails with open communication to eliminate politics
“And so when you are leading groups of people, if you want to reduce the amount of politics and people going off and, you know, forming side cliques and all this, stop having one-on-ones, have big meetings with everyone who you want to tell something and tell t…”
Ross: Groq had about 450 employees during NVIDIA deal
“About 450, but it was more difficult to manage than a typical group of 450.”
Ross: West Coast VCs act like lemmings, East Coast VCs think independently
“Well, the way I like to put it is typical, typical, not all, but typical West Coast VCs are more like Lemmings, and typical East Coast VCs all think that they're smarter than each other.”
Ross: Putting excess capital into startups no longer provides an advantage
“What's changed though is that unlike the Keynesian beauty contest where the winner is the one with the most money, in reality, there's a point at which you get enough money and you don't need more. And for the first time in history, the startups are not starve…”
Ross: GPUs excel at LLM attention while LPUs excel at applying weights
“The LPU and GPU, as mentioned, they, they're better at different parts of what's, what's called the decoder layer of an LLM. The GPU is better at the attention portion, and the LPU is better at sort of applying the weights, which is the thing that gets trained…”
Ross: Splitting LLM pre-fill and generation across different chips is a mistake
“What we realized was, and this is what most people get wrong when they're trying to do this themselves. They'll, they'll take the reading of tokens or what's called pre-fill, and they'll do that on one piece of hardware, and then they'll put the generation of …”
Ross: Porting AlphaGo to TPUs jumped its ELO by hundreds of points
“AlphaGo running on GPUs had an ELO score of about 3200. Lease it all was about 35, 35 50. And I might be getting the first digit wrong. It might be 2000 instead of 3000. But like, you know, it was more than 200. And then when put on on TPUs, it actually went t…”
Ross: AlphaGo running on GPUs never found Move 37
“In the second game, there was this famous move called move 37, which was creative. It was original. It actually wasn't completely original. It was a one in 10,000 game move. It had been in the canon of games that we're trained on. But when we went back and pla…”
Ross: GPUs are now better than Google TPUs
“So over time, these GPUs have, you know, gotten, you know, as good and better than TPUs, you know, as the career of the TPU, I have to admit, GPUs are now better.”
Ross: Faster inference compute directly makes AI models smarter
“Now bring in the LPU, and you can go deeper, faster, you can search faster, and so you can actually make a model smarter by making it faster. And so the realization is, and now that you've got this ability to reflect, think deeply, and change the outcome based…”
Ross: GitHub CEO asked Groq for LPUs amid GPU shortages
“When LLMs first started to become a thing, I remember getting a phone call from the CEO of GitHub basically saying, I need a bunch of GPUs. We've now gotten LLMs to be able to do code completion, even though, you know, we're part of Microsoft and everything yo…”
Ross: Groq Faced High Turnover Over Skepticism About Fast Inference
“There was a lot of pushback on, like, we had a lot of turnover at this point. A lot of people were leaving. This was probably three or four years ago. And so a lot of people were leaving, saying there was no point to fast inference. It wasn't going to add any …”
Ross: Stating 'I intend to' prevents pessimism and filters feedback
“And the basic idea in intentional leadership is if I ask someone, should I do something? Oh, they have opinions. Most people will be pessimistic and give you negative opinions. On the other hand, if you express intentional leadership, you say, I intend to do t…”
Ross: 80% of Groq employees took salary cuts to statutory minimums
“80% of the employees participated. And I think about half went to the statutory minimum salary by law. And remember, engineers get paid hundreds of thousands of dollars. These folks like cut their salary down to like 50, 60,000 dollars, whatever the statutory …”
Ross: Hiring requires vetting negatives rather than searching for positive traits
“The biggest flip in my hiring was when I went from looking for positives, which is what you do when you're trying to grow talent, to looking for negatives, which is what you do when you're trying to select talent.”
Ross: Marginal cost of software code is approaching zero
“What I'm seeing is that code is becoming almost free. It's, it, the marginal cost is approaching zero.”
Ross: Modern education must test question formulation over answer lookup
“When you create homework or a test for kids, if they can look up the answer online, or if they can ask AI to solve it, you haven't taught them what they need for the next stage. But if you give them a problem where they have to ask the questions and get AI to …”