Podcast InsightsPodcast InsightsPODCAST INSIGHTS
中/EN

No Priors

No Priors
Host Elad Gil, Sarah GuoOfficial Site ↗Technology & AI

Podcast Articles

8 articles
View all →

Episodes

8 episodes · newest first

Beam: The Great American Open Model with ReflectionAI Co-Founder and CEO Misha Laskin

Misha Laskin, co-founder and CEO of Reflection AI, joins Sarah Guo and Elad Gil on No Priors to talk about Beam, the company's first open model. A year of sprint took Reflection from about thirty people to around three hundred, with teams assembled across pre-training, mid-training and reinforcement learning, and its first models trained end to end. The founding logic: Laskin and his co-founder Giannis came from the Gemini reinforcement-learning team at DeepMind, bet that RL would make models agentic, and discovered a year in that all the good open models were coming from China — and that making RL work at scale requires pre-training your own model, so the lab went end to end. The training economics: Beam is a 500-billion-parameter model with 23 billion active; pre-training took 6,000 GB300s (about twelve days now, with infrastructure efficiencies), while reinforcement learning took a little over 10,000 GB300s for four weeks — more flops on RL than on pre-training, at what Reflection believes is the largest scale ever done in open source. Capex has a ceiling — hundreds of billions in, trillions unlikely — but compute efficiency compounds: roughly 7x a year with human researchers, something like 30x across systems now, and both the intelligence per flop and the revenue per flop keep rising. AlphaGo's never-stopping improvement plot is the reason Laskin got into AI, and the field has moved from can you train these things to where do you get the data. The enterprise story: companies consider open models only after ramping significant closed workloads ($100 million plus a year is not uncommon); most enterprise token consumption will come from customized systems around open models rather than customized models; on-prem is resurging on a compute shortage. At gateways like OpenRouter the token mix has flipped almost exactly from 70/30 closed-open to 70/30 open-closed, and the end state looks like operating systems, where 95%-plus of servers run open source while closed companies stay extremely valuable. On China: the open-model ecosystem is a gift with real advantages — industrial-scale distillation, effectively free copyrighted data, some state support — and a captive market; open models are Trojan horses for the infrastructure that comes with them, the 5G and Belt-and-Road pattern all over again. On safety: the default state of the world is that openness is safety — strong encryption went open after closed-side failures and birthed cybersecurity, Linus's law extends to vulnerabilities, and the science of alignment so far is whack-a-mole. The close: researcher head count settles around a hundred, pods of five to ten per capability, and a data center is what a factory used to be for a town.01:09:44 · 2026-10-09

Frontier Chips for Frontier AI Labs, with Walter Goodwin, Founder/CEO of Fractile

Walter Goodwin, founder and CEO of Fractile, joins Sarah Guo on No Priors to explain his full-stack AI chip company. Founded in the summer of 2022 on two signals — the emergence of foundation models and the AlphaGo-style argument for channeling more compute into models at test time — Fractile builds very fast inference chips for the largest models in the world. The landscape mapping: hyperscaler in-house chips (Google's TPU, Meta's MTIA, Microsoft's Maia, OpenAI's effort) ultimately ship through a relatively small number of ASIC companies with Broadcom the largest, and beneath the apparent luxury of choice the platforms share HBM, tensor cores, and TSMC packaging. Few efforts span the entire silicon stack from architecture to foundry interaction — the gap Fractile targets with about 150 people covering front-end design, physical design, backend implementation, and advanced packaging in one flexible loop. The technical bets: inference-first when some competitors' inference chips were training chips eighteen months ago; a first-two-years SRAM chip in the Groq or Cerebras mold; the late-2023 scalability worry as parameters and context lengths grew; and the pivot to DRAM — aggressively high bandwidth from the world's cheapest memory — on a platform the company says will be fully operational in the second half of next year. A faster chatbot is the faster horse; comfortably running trillion-parameter models at many thousands of tokens per second is what long-lasting agents need. The countercurrent middle: workloads move at software speed — a new model roughly every two weeks, attention mechanisms in advanced Chinese open-source models shifting every few weeks — against physical and financial limits: three-to-five-month foundry turns and three-to-five-year amortization windows, so compress the cycle and keep making subtle bets; a structurally secured three-to-six-month lead is the chip-industry equivalent of the leading model. AI enters the design loop (question assumptions, divide time limits by four) while Cadence and Synopsys keep final signoff. The bandwidth thesis: twenty years brought a million-fold compute scaling against roughly forty-fold memory bandwidth, and 25 times more bandwidth per chip than an HBM part makes MoE sparsity servable. The closing market structure: in-house chips as Nvidia-price leverage, a premium speed segment for leading labs under open-source (Kimi) pressure, and the asymmetric game that keeps labs on shared platforms — the case for independent chipmakers for decades to come.35:39 · 2026-10-02

Re-Founding Incumbents for the AI Era with Sequence Holdings Co-Founder and CEO Michael Lee

Days after announcing the $7.7 billion take-private of insurance broker Baldwin alongside Michael Dell's family office, Sequence Holdings co-founder and CEO Michael Lee joins Sarah Guo on No Priors to explain the firm the deal was built around. Sequence, founded twenty months ago after Lee's years covering AI at Lone Pine, is a perpetual holding company built on one conviction: the biggest technological shift of our lifetime rewards incumbent leaders, who already hold the brand, scale, network-effects, and regulation advantages. Pair the right incumbent with a frontier engineering team and permanent capital, the argument runs, and you can reorganize the company around what machines do best — instead of attaching "little machines" to each person on the human assembly line. The structural case walks through a Fortune 500 CEO's three failed options: do it yourself (you cannot attract restructuring talent when the firm's famous person is an investor), hire services (Accenture, McKinsey, Palantir optimize wallet share — the path to gradualism), or buy software (a beta available to everyone, selling the workflow as designed today). Only ownership aligns engineering, incentives, and horizon. The proof of work is BankSouth, a Georgia community bank that became Sequence's first investment after a service engagement and Fed and OCC approval: in about six months average consumer-loan underwriting volume fell 94 percent, commercial loans went from 30 days to 11, and second-quarter volume doubled with unchanged underwriting standards and a smaller team — a doubling Lee attributes to luck. The second half covers why brokerage — over $2 trillion in premiums, insurers rather than clients paying, roughly 90 percent retention — resists startups and why Baldwin fit; Atlas, the four-layer platform built inside the bank and expected to generalize on an 80/20 homogeneous-versus-vertical split; the contrast with fund-structured private equity and its three-year packaging reflex; and a closing underwriting stance: ideas are cheap, execution is rare, so bet on exceptional people working on hard problems in big markets — from meeting Jensen Huang in 2017 to public markets.42:44 · 2026-09-24

Building a $200M Bootstrapped Chess Empire with Chess.com CEO Erik Allebest

Chess.com co-founder and CEO Erik Allebest describes the company's growth, its financing choices, the continuing anti-cheating problem, and a view of AI as a way to support learning and product coaching rather than simply replace human skill.46:08 · 2026-08-13

Chasing Trillion-Dollar Companies, Founder Ambition, Token Budgets, & Regulatory Capture

No Priors hosts Sarah Guo and Elad Gil discuss the unusually rapid growth of recent AI companies, startup ambition and exits, compute allocation, possible technical disruption, and the tradeoff they see between AI safety rules and technological progress.39:30 · 2026-08-06

Building an Autonomous Enterprise for Real-World Services with Netic Founder Melisa Tokmak

Netic founder Melisa Tokmak describes building an AI product for essential real-world service businesses, choosing a scalable platform rather than an acquisition roll-up, and viewing the next phase of AI adoption through operations, measurable ROI, and human-delivered services.34:30 · 2026-07-31

Building an Autonomous Delivery Experience with DoorDash Co-Founders Andy Fang and Stanley Tang

DoorDash co-founders Andy Fang and Stanley Tang describe a single strategy connecting conversational commerce, physical delivery automation, and company-wide AI adoption: begin with a concrete customer workflow, deploy into the real operating environment, and let observed constraints determine the product. Ask DoorDash is presented as an early agentic interface that changes discovery and basket construction, while DoorDash Dot is presented as a purpose-built response to the mismatch between sidewalk robots and passenger robotaxis. The episode's deeper argument is that physical-world autonomy is not primarily a model demonstration. It requires dispatch, merchant integration, first-and-last-100-foot knowledge, fleet maintenance, manufacturing, and the ability to route each delivery to an appropriate human or machine modality. The final section applies the same realism to enterprise AI: benchmarks and sanitized tasks are insufficient unless model costs and performance are measured against actual work.49:20 · 2026-07-23

Travel Through the Lens of AI with with Booking.com CEO Glenn Fogel

Booking Holdings CEO Glenn Fogel uses Priceline's dot-com collapse and recovery to argue for restraint amid the AI investment boom: transformative technology can create enormous value while still destroying most companies built around it. Applied to travel, his position is neither defensive nor utopian. AI is valuable when it reduces planning friction, coordinates complex itineraries, responds when disruptions cascade, and improves outcomes for travelers and travel partners. The episode repeatedly returns to scale and economics. Penny demonstrates what an agentic travel product can do, but Fogel stresses that absolute adoption, token costs, model selection, conversion, loyalty, customer-service outcomes, and lifetime value still determine whether deployment creates value. Booking's inventory, partner operations, regulatory capacity, and global marketplace are advantages rather than permanent protection. The final section widens the lens to labor: technological gains are real, but job destruction, job creation, and worker retraining do not necessarily occur at matching speeds.41:05 · 2026-07-09