Discover
The Compute 100
The Compute 100
Author: The Compute 100
Subscribed: 1Played: 12Subscribe
Share
© The Compute 100
Description
The Compute 100 is the home for conversations about compute.
The Compute 100 highlights 100 companies building across the semiconductor and compute value chain — 50 public, 50 private. It spans chip design, fabrication, advanced packaging, memory, power delivery, networking, and the software stack powering AI workloads. This podcast highlights conversations with the operators, builders, and investors shaping the new compute-centric economy.
The podcast is hosted by Brian Schechter and Gaby Lorenzi, early stage compute investors at Primary.
The Compute 100 highlights 100 companies building across the semiconductor and compute value chain — 50 public, 50 private. It spans chip design, fabrication, advanced packaging, memory, power delivery, networking, and the software stack powering AI workloads. This podcast highlights conversations with the operators, builders, and investors shaping the new compute-centric economy.
The podcast is hosted by Brian Schechter and Gaby Lorenzi, early stage compute investors at Primary.
16 Episodes
Reverse
Today’s guest is Roman Chernin, Co-Founder of Nebius.Roman has been at Nebius since day 1. Since being on the journey from Yandex, to launching Nebius just over 2 years ago, the company has been on a historic rise as a dominant player in the neocloud market, known for training and inference offering, vertical integration of their hardware stack, and selling compute to some of the biggest buyers in the world.Few people question demand for AI compute, which shifts attention to whether a neocloud can finance its buildout and own more of the customer experience. We get into how Nebius builds in-house, spreads projects across many sites and how they’re thinking about funding new capacity.From financing to GPU reliability, we discuss:- Why building data centers, racks and software in-house changes Nebius’ costs- How short-term capacity sales and prepayments help fund new buildouts- Where demand is coming from across labs, AI-native products and enterprises- Why agentic workloads are driving demand for CPU compute and sandboxes- What makes GPU capacity reliable at scale, from failing nodes to storage speedTimestamps:(00:00) Introduction(01:04) Market swings and the DeepSeek stock drop(04:59) Vertical integration and financing discipline(08:50) From Yandex to Nebius(12:32) Data-center delays and revenue worries(17:02) Where Nebius builds its data centers(19:47) Financing the buildout and the Meta backstop(24:31) Short-term capacity and customer prepayments(29:30) Where AI demand is coming from(38:02) Nebius’ product layers and Roman’s role(40:38) Agentic search and open models(45:05) Why agents are driving CPU demand(49:22) One compute pool for training and inference(55:10) What makes compute high quality(01:02:09) Customer advisory board and the compute rebate(01:08:09) Culture, hiring and an underdog mindsetThe Compute 100 is led by Brian Schechter and Gaby Lorenzi, compute-focused investors at Primary. Primary is a pre-seed and seed-stage venture firm that backs founders building across markets including compute, industrials, healthcare and vertical AI. With $1.6B AUM and a 60+ person operating team, Primary delivers unparalleled support to teams across recruiting, finance, GTM, and brand. Primary’s compute portfolio includes Etched, The Biological Computing Company, Haiqu, and Atero (acquired by Crusoe).Follow us on X and check out our daily newsletter.This podcast is provided for informational purposes only and does not constitute an offer to sell or a solicitation of an offer to buy any security. Nothing discussed in this podcast constitutes investment, legal, tax, or accounting advice, or a recommendation to make any investment. Any views expressed are as of the date of recording, are subject to change, and may include forward-looking statements that involve risks and uncertainties.
Today’s guest is Dan Fu, SVP of Research at Together AI.A co-creator of FlashAttention, Dan now leads research at a company that provides the compute and software to train and serve AI models.We get into the race to bring new open models to market: matching the developer’s quality benchmarks, adapting to changing architectures and deciding where optimization effort will pay off. Dan explains how those research choices become commercial decisions about cost, capacity and the customers Together can serve.From GPU instructions to the economics of open models, we discuss:What CUDA, kernels and compilers each doHow mixture of experts and new attention designs change hardware requirementsWhy FlashAttention came from rethinking how data moves through a GPUHow AI is changing the work of kernel engineersHow inference revenue helps fund open model developmentChoosing between owned GPUs and leased compute as customer demand growsTimestamps:(00:00) Introduction(01:04) From AI models to CUDA kernels(11:38) Launching new models at Together(22:36) How model architectures are changing(27:42) Comparing open and closed models(31:57) How AI is changing kernel engineering(38:01) From systems research to FlashAttention(47:11) Using AI to improve inference(48:16) Together’s origins and the business of open models(51:17) Owning versus leasing compute(52:55) Building the research team and evaluating new chips(54:56) Frontier labs, open models and partnerships(57:16) Inference profits and funding the next model(01:00:13) Academic research and developing new talent(01:02:47) Robotics and AI in the physical worldThe Compute 100 is led by Brian Schechter and Gaby Lorenzi, compute-focused investors at Primary. Primary is a pre-seed and seed-stage venture firm that backs founders building across markets including compute, industrials, healthcare and vertical AI. With $1.6B AUM and a 60+ person operating team, Primary delivers unparalleled support to teams across recruiting, finance, GTM, and brand. Primary’s compute portfolio includes Etched, The Biological Computing Company, Haiqu, and Atero (acquired by Crusoe).Follow us on X and check out our daily newsletter.This podcast is provided for informational purposes only and does not constitute an offer to sell or a solicitation of an offer to buy any security. Nothing discussed in this podcast constitutes investment, legal, tax, or accounting advice, or a recommendation to make any investment. Any views expressed are as of the date of recording, are subject to change, and may include forward-looking statements that involve risks and uncertainties.
Today’s guest is Timothy Arcuri, Managing Director at UBS.Timothy leads UBS' Equity Research efforts in semiconductors and semiconductor equipment in the US region. A former process engineer, Tim began his Wall Street career in 1998 and has spent nearly three decades studying chipmakers and their supply chains.Timothy has been covering the semiconductor landscape for over 20 years. His expertise is unmatched.Across the compute stack, we discuss:- The power and revenue needed to support semiconductor growth expectations- Who will pay a premium for faster inference- How hyperscalers balance Broadcom’s full-service model with greater control over chip design and memory procurement- What Cerebras’s OpenAI deal means for its cloud and hardware business model- Where equipment, packaging and materials create openings for startups- Political backlash and the industry’s dependence on a small number of customersTimestamps:0:00 Introduction1:13 Can power keep up with semiconductor expectations?7:55 From process engineering to Wall Street12:40 Measuring AI demand and pricing faster inference18:44 How AI is changing semiconductor research21:12 NVIDIA, AMD and Intel24:31 Custom chips, Broadcom and MediaTek28:57 Inference startups and Cerebras’s business model33:22 Micron, HBM and the memory cycle38:54 What long-term memory contracts change45:50 Equipment, packaging and fab economics53:56 Where new compute companies can compete1:02:27 Cobots, humanoids and demand for chips1:03:57 Political backlash and customer concentration1:08:07 NVIDIA’s installed base and what gets cut in a downturnThe Compute 100 is led by Brian Schechter and Gaby Lorenzi, compute-focused investors at Primary. Primary is a pre-seed and seed-stage venture firm that backs founders building across markets including compute, industrials, healthcare and vertical AI. With $1.6B AUM and a 60+ person operating team, Primary delivers unparalleled support to teams across recruiting, finance, GTM, and brand. Primary’s compute portfolio includes Etched, The Biological Computing Company, Haiqu, and Atero (acquired by Crusoe).Follow us on X and check out our daily newsletter.This podcast is provided for informational purposes only and does not constitute an offer to sell or a solicitation of an offer to buy any security. Nothing discussed in this podcast constitutes investment, legal, tax, or accounting advice, or a recommendation to make any investment. Any views expressed are as of the date of recording, are subject to change, and may include forward-looking statements that involve risks and uncertainties.
Today’s guests are Barun Kar, CEO, and Rajiv Khemani, Executive Chairman at Upscale AI.Barun and Rajiv are multi-time founders in networking silicon. Between them they helped build Cavium, Innovium, Palo Alto Networks and Juniper’s MX series before co-founding Auradine, where they incubated Upscale AI before spinning it out. Upscale builds the switch silicon, systems and software that connect AI chips to each other using open standards.We dig into why AI clusters need a network built around memory reads and writes between chips, not the packet switching that cloud networks were designed for, and why Barun and Rajiv chose open standards over a proprietary fabric like NVLink. They explain how a networking startup works with hyperscalers on their custom chips, why it sells complete systems to new clouds and enterprises, and how the AI buildout could make networking a trillion dollar market.From switch silicon to supply chains, we discuss:How scale-up and scale-out networks workWhere copper, optics and optical switches fitWhy open standards matter for AI chipsWhat a network operating system doesHow Upscale secures its supply chainBuilding culture while scaling to 300 peopleTimestamps:0:00 Introduction1:05 Why AI needs a new network5:21 Why cloud-era chips fall short for AI9:05 Scale-up and scale-out inside an AI rack14:47 The Auradine spin-out and Upscale’s founding story19:30 Memory semantics versus packet switching26:14 From Cavium and Palo Alto Networks to Upscale30:50 Competition and the system-level advantage34:19 What a network operating system does35:24 Optics and optical switching40:33 Selling to hyperscalers, new clouds and enterprises46:04 How big AI networking could get49:04 Securing supply across the chain53:31 Building at the forefront of AI56:17 Beyond networking: what comes next1:02:23 Building culture in a fast-growing companyThe Compute 100 is led by Brian Schechter and Gaby Lorenzi, compute-focused investors at Primary. Primary is a pre-seed and seed-stage venture firm that backs founders building across markets including compute, industrials, healthcare and vertical AI. With $1.6B AUM and a 60+ person operating team, Primary delivers unparalleled support to teams across recruiting, finance, GTM, and brand. Primary’s compute portfolio includes Etched, The Biological Computing Company, Haiqu, and Atero (acquired by Crusoe).Follow us on X and check out our daily newsletter.This podcast is provided for informational purposes only and does not constitute an offer to sell or a solicitation of an offer to buy any security. Nothing discussed in this podcast constitutes investment, legal, tax, or accounting advice, or a recommendation to make any investment. Any views expressed are as of the date of recording, are subject to change, and may include forward-looking statements that involve risks and uncertainties.
Today’s guest is Jascha Achterberg, co-founder and CTO of Callosum. Jascha started with his Cambridge PhD collaborator Danyal Akarca (CEO), London-based Callosum is building infrastructure that matches complex AI workloads to specialized models and chips. The company has raised a $100 million seed round led by Atomico, with significant participation from Plural, DCVC, and the UK Sovereign AI Fund.Jascha’s bet is that the future of compute is heterogeneous: different chips working together on different parts of a problem. We’ll dig into the commercialization opportunity: discover where each architecture excels, help chipmakers find production workloads, and let enterprises adopt new compute without replacing the stack they already trust. It is all in pursuit of faster, cheaper, and more capable computing systems, which is the biggest prize in infrastructure today. Beyond the multichip future, we discuss: What neuroscience taught Jascha about designing computing systemsHow chipmakers discover what their hardware is best atWhere thermodynamic, biological, and quantum computing could fitHiring people who welcome having their ideas challengedBuilding an AI infrastructure company in LondonTimestamps:0:00 Introduction2:55 What heterogeneous computing actually means11:02 From PhD research to Callosum18:36 Turning heterogeneous compute into a product23:24 Sourcing compute and working with chipmakers28:53 Matching workloads to the right hardware34:24 What heterogeneous computing can do today38:29 The next waves of chips48:02 Building the team51:33 Why build in London?56:11 Helping enterprises upgrade their compute stacks1:01:04 Fundraising and building momentum1:03:40 Where specialization wins, and where it doesn’tAnd a few links mentioned in the show: Callosum’s recent results on heterogeneous inference deployments: https://www.callosum.com/blog/programmable-heterogeneityBlog post on Callosum’s theoretical research on heterogeneity across computing, economic, and biological systems: https://www.callosum.com/blog/the-principle-of-maximum-heterogeneityThe Compute 100 is led by Brian Schechter and Gaby Lorenzi, compute-focused investors at Primary. Primary is a pre-seed and seed-stage venture firm that backs founders building across markets including compute, industrials, healthcare and vertical AI. With $1.6B AUM and a 60+ person operating team, Primary delivers unparalleled support to teams across recruiting, finance, GTM, and brand. Primary’s compute portfolio includes Etched, The Biological Computing Company, Haiqu, and Atero (acquired by Crusoe).The Compute 100 is led by Brian Schechter and Gaby Lorenzi, compute-focused investors at Primary. Primary is a pre-seed and seed-stage venture firm that backs founders building across markets including compute, industrials, healthcare and vertical AI. With $1.6B AUM and a 60+ person operating team, Primary delivers unparalleled support to teams across recruiting, finance, GTM, and brand. Primary’s compute portfolio includes Etched, The Biological Computing Company, Haiqu, and Atero (acquired by Crusoe).Follow us on X and check out our daily newsletter.This podcast is provided for informational purposes only and does not constitute an offer to sell or a solicitation of an offer to buy any security. Nothing discussed in this podcast constitutes investment, legal, tax, or accounting advice, or a recommendation to make any investment. Any views expressed are as of the date of recording, are subject to change, and may include forward-looking statements that involve risks and uncertainties.








