AGENDA

SOSV Founder and Managing General Partner Sean O’Sullivan opening remarks at the SOSV Deep Tech Summit.

Genomic medicine is moving fast, and Nobel Laureate Dr. Jennifer Doudna is very much at the forefront. In September, her UC Berkeley lab published the discovery of VIPR, a potentially new powerful gene-targeting system that might one day rival CRISPR, and in July, her company Scribe Therapeutics went public, promising a CRISPR-based therapy designed to lower LDL cholesterol.

In a very special fireside chat, Dr. Doudna and SOSV General Partner Stephen Chambers will discuss the latest CRISPR advances and where genetic medicine will lead in the years ahead.

Can AI fundamentally change how science gets done? With breakthroughs like DeepMinds’ AlphaGenome Atlas, we’re getting more glimpses of how AI can accelerate science. But, given the extraordinary vastness of  fields like physics, chemistry, and biology versus finite human abilities to experiment and make discoveries, what big steps forward can AI technologies provide, and how will that unfold? Can cancer really be “cured” in the coming decade, as some AI champions predict? What would that take?

Insitro’s Daphne Koller brought machine learning to drug discovery. Now Sam Rodriques is building autonomous AI scientists through Edison Scientific and FutureHouse.

This panel moderated by famed science and pharma writer Derek Lowe will explore the future of AI in science.

What happens when you apply AI or machine learning across the entire value chain in a physical AI startup? Variation on those technologies end up not only driving product and process, but also product and business decisions in ways that founders could only have imagined a few years ago.

Avalo.ai, an AI-powered plant evolution platform, uses interpretable ML to accelerate crop breeding for climate resilience. Vertical Oceans relies on AI to operate vertical aquaculture, delivering delicious shrimp at market prices. Allozymes harnesses proprietary microfluidics and machine learning to design custom enzymes for pharma, food, and chemicals. But that’s just the start.

This panel brings together three founders rewriting what’s possible when silicon meets biology, and building what they call the “physical AI stack.”

Startup gurus Steve Blank and Eric Ries invented the “lean startup” playbook that powered the past generation’s startup super cycle. In recent posts, Blank has argues that AI has changed startups so profoundly that even the classic Minimum Viable Product (MVP) concept needs an overhaul. Now that almost anyone can build a polished prototype in days (or hours), the hard part is knowing what to build, for whom, and how to create a defensible business. As a result, Blank has upended the curriculum for his long-running Lean Launchpad course at Stanford’s Graduate School of Business. Who better to interview Blank about the upheaval than Eric Ries, Blank’s friend, collaborator, and author of the 2011 startup bible, The Lean Startup. Eric has also just published another blockbuster, Incorruptible, which  focuses on governance, incentives, and how great companies can be designed to endure.

Brain-computer interfaces promise something straight out of sci-fi: thought to speech, control of computers and robots just by thinking, even new senses we were never born with. But large language models learned from words, a latecomer in brain evolution. What could AI learn from the brain itself? Synchron is building a foundation model trained directly on human neural activity. General Sense reads what detection dogs smell in real time, giving AI a sense it has never had. Meta’s Jean-Rémi King uses AI to decode typed sentences from brain recordings taken entirely outside the skull. Bing Brunton models how brains and bodies work together, and helped write a roadmap for AI built on those principles. MIT Technology Review’s Antonio Regalado asks what AI gets when it stops reading our words and starts reading our brains.

Zachary Bogue and Matt Ocko launched DCVC back in 2010, when few venture capitalists shared their thesis—that venture could reap huge wins by applying computational approaches to problems in the physical world, often in industries far afield from Silicon Valley. They called it “deep tech,” and the term stuck. Today, DCVC has more than $4 billion under management and a portfolio that spans AI, exascale computing, energy and climate, quantum, materials science, robotics, space, water, biology, defense and security, and more. The rapid rise of physical AI-based technologies and an attendant venture-backed investment surge are only the latest expression of how on-target Ocko’s and Bogue’s thesis was almost two decades ago. TechCrunch’s editor-in-chief, Connie Loizos, will discuss with the co-founders what they have learned along the way, what they think of the present moment, and where their investment strategy will go from here.

The Hormuz choked. Russian refineries blitzed. Houthi pipeline attacks. Diesel’s soaring prices. Data center demand. China’s solar advances.

When the Trump administration pulled most of the Biden IRA funding last year, the climate tech scene reeled. And what no one saw coming were the even greater upheavals this year. Alarming, yes, but the new dynamics are creating favorable conditions for renewable energy – including nuclear, geothermal, and solar – as grids around the world race to adapt to oil and gas supply challenges and hyper-scalers’ vast energy demands.

Deep tech doesn’t advance in a vacuum, so we asked three of the world’s top thinkers on geopolitics, global markets, and energy to discuss the interplay of all those factors on energy markets and the impact on renewable energy.

Climate tech investing in 1H 2026 roared back to levels not seen since 2022, but the drivers would have been unimaginable four years ago. Energy investment driven by the data center build-out drove was the biggest driver of the 55% Y/Y surge according to Currence.AI, with strong funding rounds in nuclear companies, but also climate adaptation and built environment. Our panel of long-haul, deeply expert VCs from DCVC, At One, and Khosla Ventures weigh in on where they see the current wave taking climate tech and which sectors are ripe for growth.

The Biden Administration’s 2022 IRA set aside $369 billion to underwrite climate initiatives, which in turn encouraged a boom in venture-backed climate investment across energy, cement, steel, and many other carbon-intensive categories. When the Trump Administration pulled back most of that funding, hundreds of companies lost grants and loans that were baked into their business plans. Pivoting to harsher commercial realities was a monumental challenge, but these three firms pulled it off. The leaders of Rondo Energy (industrial heat batteries), Sublime Systems (low-carbon cement), and Brimstone (industrial materials) successfully escaped the IRA’s implosion and are rolling toward commercial success. Their founders will talk about what it took to make the shift.

SOSV Founder and Managing General Partner Sean O’Sullivan kicks off the second day of the Summit.

Like many science-fiction writers, Neal Stephenson has been weaving AI-themes into his novels from the start. In Snow Crash (1992) he previewed conversational AI interfaces, multimodal retrieval-augmented generation (RAG), and agentic assistants, and in Diamond Age (1995) imagines an interactive, nanotech-powered book designed to tutor a child into becoming an autonomous, critical thinker, with an important qualification. The experience requires a human behind an avatar to supply the subtle emotional responsiveness that code alone cannot achieve. In Fall; or, Dodge in Hell (2009), Stephenson builds out a societal hellscape brought on by bot swarms that results in a badly bifurcated society. You could say, Stephenson has previewed things that have arrived and some that may be around the corner. Jason Pontin, former editor of MIT Technology Review and now general partner at DCVC, will talk with Stephenson about how he sizes up our AI present and future.

Stephenson’s latest novel D: Heavy Water comes out in mid-October.

MIT professor Russ Tedrake has spent decades working on one of robotics’ hardest challenges: building machines that can learn, adapt, and act autonomously in the physical world. After co-founding Walden Robotics and raising $300 million at a $1.1 billion valuation, he is putting machine intelligence to the test in the real world.

Can robots learn from humans and from their own deployments, and ultimately handle the complexity and variability of real-world work?

For decades, the physical world – factories, energy systems, robotics, biotech – was often considered too capital intensive and slow moving for venture-scale returns. AI is helping rewrite that equation.

Investment is surging. Physical AI startups raised $47 billion globally in the first half of 2026, nearly 4x the previous six months. And some of deep tech’s leading investors are raising billions to pursue the opportunity: Eclipse recently raised $1.3 billion, Lux Capital $1.5 billion, and Playground Global $475 million.

What has changed, and where are the biggest opportunities now? These investors will discuss where AI meets the physical world, what can produce venture-scale returns, and where the hype may be running ahead of reality.

Seasats removes the captain. SiFly removes the pilot. Bot Auto removes the driver.

Three Series A startups are each building fully autonomous systems for radically different customers and environments—maritime logistics, long-endurance delivery drones, Level 4 freight trucks.

This session brings together founders betting that autonomy is foundation on which they can disrupt these crowded, deeply established categories of transportation.  We will discuss the extent to which their technology and mission challenges are similar or different, and how their markets are shaping their products and business trajectory.

The physical, real-world edge for AI applications poses unique demands: cameras, microphones, and many other sensors generate continuous streams of data that must be interpreted in milliseconds, sometimes with life-or-death consequences. The conventional cloud compute model places the compute that can make sense from the signal far away. Will that model hold up?

Kenyi Technologies and Qualcomm each offer hardware solutions to the challenge, but interpret the problem differently. Archetype AI brings multimodal foundation models into the conversation: can software-centric approaches handle the speed and complexity that real-world sensors demand?

This panel explores how startups and chipmakers approach the latency, power, and accuracy tradeoffs that make or break edge deployment.

Industrial robotics is entering a new era as advances in AI, vision and robotics expand what machines can do. VCs investment is surging, and so are startup valuations in the category. Agility Robotics is preparing to go public at a $2.5B valuation, Standard Bots raised $200M this year at a $1B valuation, and Inbolt has raised over $25M while deploying across 100+ factories.

These three robotics experts approach this future differently. Damion Shelton, who recently took on a role at Carnegie Mellon University, co-founded Agility Robotics, where he was a pioneer bipedal humanoids. Standard Bots’ Leif Jentoft is building AI-native industrial robots from the ground up. Inbolt’s Albane Dersy adds AI-powered vision to existing robots.

We’ll dig into how the latest wave of physical AI advances are changing the outlook for industrial automation. Are we approaching an era of general-purpose bipedal robots on factory floors, or will specialized machines dominate? How close are we to solving the human-robot interaction challenges?

Physical AI is attracting a new mix of investors, from global technology and industrial giants to specialist deep-tech funds.

Samsung, a $1 trillion+ technology conglomerate, TDK, a century-old leader in advanced electronics and materials, and Physical World, a new specialist deep-tech VC bring very different investment models, technical expertise, industrial networks and paths to market.

What are they looking for as investment in Physical AI accelerates? This panel explores how corporate and specialist investors evaluate deep-tech startups, where they see the biggest opportunities, and what they can offer founders beyond capital, from technical expertise and manufacturing to customers, partnerships and global scale.

For forty years, RAS was the textbook cancer target no drug could touch. In August, the FDA approved daraxonrasib, a RAS inhibitor that nearly doubled median survival in metastatic pancreatic cancer. Its chemistry traces back to Warp Drive Bio, which Greg Verdine founded and Revolution Medicines acquired in 2018. Verdine is now CEO of LifeMine Therapeutics, which mines fungal genomes for new drugs, and founder of Parabilis Medicines, which went public in June. He has made a career of targets the field wrote off. In this conversation, he’ll explore what impossible targets could fall next, and how long it will take.

Bringing a new medicine to patients takes 10 years or more, and the delays show up at every step. So do the efforts to cut them, across both the drug pipeline and production.

Ten63 Therapeutics CEO Marcel Frenkel runs an AI-driven discovery engine that searches 19.5 million compounds a second. Prellis Biologics CEO Mike Nohaile grows fully human antibodies in lab-built lymph node organoids, speeding discovery and development while generating the human data AI models need. That approach landed a deal with Eli Lilly last year. Prolific Machines CEO Deniz Kent uses light to control exactly when and how much drug the cells in a bioreactor produce. In May his team hit 21 grams of antibody per liter in a single production run, more than double the industry norm.

SOSV General Partner Po Bronson asks each of them: how fast can we go, and what will it take?

Vaccine platforms are racing ahead, from viral vectors and mRNA to self-amplifying RNA and vaccines built to cover whole families of viruses at once. This session weighs their promise and their challenges with three world experts.

Dan Barouch, Director of the Center for Virology and Vaccine Research at Beth Israel Deaconess, took his COVID vaccine from the lab to human trials in about six months. Seth Berkley co-founded the global effort that got 2 billion doses to people, and now advises on vaccine technology. Jon Cohen has covered the field for Science for decades and recently wrote Planning Miracles: How to Prevent Future Pandemics. Together, they’ll explore how these platforms could help us respond better and faster to the next pandemic – or prevent it in the first place.

Jim Collins helped launch synthetic biology by showing that living cells could be programmed like circuits. More than 25 years later, gene circuits are in human cancer trials. Now Collins, the Termeer Professor of Medical Engineering and Science at MIT, is taking on another huge problem: using AI to discover antibiotics faster than bacteria can evolve resistance.

In June, his lab reported using an AI model that found dozens of compounds active against drug-resistant gonorrhea. It outperformed a large language model and unearthed chemistry unlike any known antibiotic.

Jessica Hamzelou of MIT Technology Review asks Collins what happens when we can program biology at both ends: engineering living cells and using AI to invent the medicines that act on them.

SOSV Founder and Managing General Partner Sean O’Sullivan launches the third and final day of the SOSV Deep Tech Summit.

In June, David Sinclair’s Life Biosciences started clinical trials for ER-100 to see if it can restore vision to people with optic nerve damage, just as a precursor did in mice, six years ago. Built on work at Sinclair’s lab at Harvard Medical School on partial epigenetic reprogramming, ER-100 switches on three genes that reset a cell to a younger, pre-disease state.

The trial, the first of its kind at the FDA, marks a major test for Sinclair’s Information Theory of Aging, which holds that aging is driven by the gradual loss of the epigenetic instructions that tell cells how to use their DNA. If Sinclair is right, ER-100 may go down as one of the greatest breakthroughs yet in the burgeoning science of human longevity. Dr. Sinclair and SOSV General Partner Mohan Iyer will talk in depth about the ER-100 project and more in a very special opportunity to hear from one of the world’s foremost researchers on aging and longevity.

Of the 100 generic drugs Americans use most, 83 have no U.S. source for their active ingredient. That’s a national security problem. This panel brings together three people attacking different parts of it.

Michelle Rozo invests for the national security community and is vice chair of the National Security Commission on Emerging Biotechnology; she is also a past advisor to the White House on biotechnology and national security. Alex Oshmyansky, co-founder and CEO of Mark Cuban Cost Plus Drug Company, runs a robotic plant in Dallas that turns ingredients into finished drugs, and was picked by HHS and DARPA to test AI-guided production. Elizabeth Onderko, co-founder and CEO of SOSV-backed Capra Biosciences, grows drug ingredients in bioreactors instead of building them from imported chemicals, with about $17 million from HHS to bring seven critical pharmaceutical ingredients home.

SOSV’s Pae Wu asks what it will take to rebuild a U.S. drug supply chain that can withstand the next crisis.

Until recently, serious ailments in women like Endometriosis could take seven years to diagnose. For decades, women’s health has relied on blunt tools: symptoms instead of biomarkers, animal models that don’t always translate to human disease, and too few targeted treatments. That is starting to change, with new nonhormonal menopause drugs reaching patients and new ways to diagnose and study women’s conditions close behind.

Jocelyn Wessels, co-founder and chief scientific officer of Afynia, developed a blood test that reads microRNA to detect endometriosis without surgery. Morgan Stanton, founder and CEO of Opal Therapeutics, grows tiny models of the uterus from menstrual blood so drugs can be tested on human tissue. Jenica Patterson led ARPA-H’s $113 million Sprint for Women’s Health and now runs the Milken Institute’s Women’s Health Network.

SOSV NY Chief Science Officer Sabriya Stukes asks what the science needs next to close the gap.

Thanks to rapid advances in monitoring and wearable tech, just about anyone’s genome, proteins, metabolites, and microbes can be tracked in near real time. The promise is catching disease earlier. The problem is that medicine is struggling to make use of the data bonanza.

Michael Snyder, chair of genetics at Stanford, has been called medicine’s most-measured man and pioneered using wearables to detect illness early. Sahan Ranamukhaarachchi, CEO of Proton Intelligence, is building a wearable that tracks potassium continuously, an indication critical to heart health in kidney patients. Shebani Sethi, who founded Stanford’s Metabolic Psychiatry Clinic and now runs Metabolic Psychiatry Labs, uses metabolic data to rethink treatment for serious mental illness.

SOSV General Partner Mohan Iyer asks when knowing more actually changes care.

For two decades, John Rogers has been making electronics soft enough to live on the body, and lately, inside it. Last year his lab unveiled a pacemaker smaller than a grain of rice that a patch on the chest switches on with pulses of light, and that dissolves once the heart no longer needs it.

In May, his team debuted a bandage-like sensor that reads hidden stress in babies too young to say what’s wrong. Rogers, who directs Northwestern’s Querrey Simpson Institute for Bioelectronics, joins IEEE Spectrum’s Eliza Strickland to talk about engineering for every environment in and on the body, from the skin to devices designed to disappear.

In June this year, Gigascale Founder and former META CTO Mike Schroepfer raised a $250M institutional fund to exploit what he sees as “a revolution in the hardware infrastructure in the physical world happening at the same time that there’s this massive increase in demand from data centers, from onshoring, from electrification….a perfect storm of disruption.” (Bloomberg). In conversation with Currence co-founder Kim Zou, Schroepfer will talk about his favorite sectors and companies, as well as the future of climate tech and its surprising resurgence, at least in the energy field.

Solar is the greatest success story in renewable energy and what’s striking is how much more solar innovation is in the works. This panel captures three very different facets of that innovation, ranging from Reflect Orbital‘s plan to she sunlight on solar fields at night, Planted Solar‘s promise to get megawatts of solar to the grid in “months, not years,” CPTI‘s application of revolutionary  perovskite tech to bring solar to almost any device or surface.  

Nuclear power supplies nearly 10% of global electricity today, but as demand climbs, some are looking beyond the 20th-century blueprint. X-energy launched a $1B IPO earlier this year to support rollout of its rapidly deployable, SMR tech, modular reactors that employ X-energy’s meltdown-proof TRISO-X fuel architecture.  X-energy is partnering Amazon’s AWS to provide 5 GW of power by 2039 to data centers and the grid. Kairos Power is innovating with the a fluoride salt-cooled high-temperature reactor that eliminates the high-pressure requirements of conventional reactors, and is partnered with Google to deliver power from SMR reactors in the 2030s. Our panelists will cover how new technology could usher in nuclear’s next act, and what challenges they have overcome as well as those still ahead of them.

Of the many competitors in the race to deliver fusion-based energy, Helion has the most ambitious time-table. The firm plans to deliver partner Microsoft 50 MW in 2028, years ahead of the rest. Helion raised a fresh $465M this year ($1.5B in total) and is valued at $15.5B so investors are clearly bullish. Helion’s technology, like all of its competitors, is not fully proven, but the R&D facility in Everett, Washington, is up and running and the first commercial plant, In Orion, Washington, is permitted and under construction. Will Helion flip the switch in 2028?