absorb.md

20VC with Harry Stebbings

Chronological feed of everything captured from 20VC with Harry Stebbings.

Navigating the AI-Driven Market: NVIDIA Dominance, Layoffs, and Redefined Skillsets

NVIDIA is experiencing unprecedented growth, driven by massive capex investment set to continue for years, despite some market skepticism regarding the sustainability of their trillion-dollar revenue projections. Concurrently, a significant trend of layoffs in tech giants like Atlassian and Meta indicates a strategic shift, where companies are reallocating resources from human capital to compute, driven by AI efficiencies. This shift necessitates a reimagined workforce, emphasizing "AI fluency" and agent deployment expertise over traditional technical skills.

Venture Capital in 2025: AI Gold Rush Meets Liquidity Drought in a Low-Low Market

The current VC environment is uniquely punishing: deal valuations are at peak levels while exit channels (IPOs and M&A) remain effectively closed, creating a "low-low" quadrant where it's simultaneously expensive to deploy and nearly impossible to return capital. The AI category is experiencing a gold rush dynamic broader and faster than any prior cycle, reaching high schoolers and non-technical founders, but the flood of capital into crowded categories is eroding investor returns even when winners emerge. Sophisticated allocators are debating whether the liquidity drought is a temporary dislocation or a permanent structural shift in how companies mature, with the answer having existential implications for venture as an asset class. IRR is increasingly seen as the least controllable fund metric — picking and entry valuation matter more — while secondaries and disciplined portfolio construction are emerging as the primary tools for generating DPI in a no-exit environment.

AI Agents Are Crossing the "Part of the Team" Threshold — And Venture Capital Isn't Ready

The 20VC panel argues that 2025 marks the inflection point where AI tools have graduated from productivity co-pilots to autonomous team members capable of completing high-value tasks end-to-end — a qualitative shift exemplified by Replit's V3 agent and Gamma's dynamic collateral generation. This transition is compressing moat-building timelines dramatically: clones that once took incumbents 3 years to ship now arrive in 30 days, rendering early-stage defensibility arguments largely obsolete. The panel concludes that seed-stage venture is fundamentally broken for non-AI-native companies, with capital concentrating almost entirely in top-decile growers, while the only durable AI investment theses are either attaching to hyperscaler compute budgets or directly replacing human labor. Simultaneously, the pace of model improvement means any framework older than 6 months is likely already wrong, creating structural stress even for top-tier firms like Sequoia.

2025 VC Year-in-Review: IPO Pipeline, AI Valuation Ceilings, and the Looming Tech Backlash

In a 20VC "Big Fat Quiz of the Year" roundtable, Jason Calacanis, Rory O'Driscoll, and Harry Stebbings assess 2025's standout founders, funds, and companies while projecting 2026. The panel converges on Dario Amodei (Anthropic) as founder of the year, crediting Claude 3.5/3.7/4 as the foundational model enabling the entire vibe-coding ecosystem. The most structurally significant observation is that venture now has "no ceiling" on exit valuations — a regime shift that invalidates older portfolio construction math. The panel also warns that any uptick in unemployment, regardless of cause, will trigger a disproportionate AI backlash because AI executives have already "confessed to the crime" by publicly predicting job displacement.

Gokul Rajaram’s Enduring Software Moats

Gokul Rajaram, a seasoned operator and investor, outlines eight modes that contribute to the durability and defensibility of software companies, categorizing them as data, workflow, regulatory, distribution, ecosystem, network, physical infrastructure, and scale. These modes, individually insufficient, collectively provide a robust framework for assessing a company’s long-term viability, particularly in the context of increasing software commoditization and competitive pressures. He stresses that strong, multi-product offerings that tackle entire solution stacks are necessary for companies to thrive.

Deconstructing the PLG-to-Enterprise Sales Transition: Figma's First-Principles GTM

Figma has abandoned traditional SaaS GTM structures (CS/SDR) in favor of a first-principles approach that blends PLG with a prescriptive, outbound expansion motion. The organization prioritizes behavioral competencies and strategic value-delivery over rigid quota attainment, treating sales as an educational process rather than a transactional one. This shift reflects a transition from simple 'order taking' in early PLG stages to complex, multi-stakeholder strategic selling as the product matures.

Anduril: Navigating the Complexities of Defense Tech and Government Contracts

Anduril's president discusses the nuances of building a successful defense technology company, emphasizing the need for broad market reach, strategic capital allocation, and a deep understanding of government procurement. The company leverages a platform approach to develop fundamental technologies applicable across various defense challenges, allowing for rapid iteration and deployment. Success in defense requires navigating complex regulatory environments, anticipating future military needs, and fostering strong relationships within the government.

Navigating the AI Investment Landscape: Market Dynamics and Strategic Shifts

The AI market is experiencing significant shifts, with Anthropic challenging OpenAI's enterprise dominance, evidenced by recent spending trends. The broader venture capital landscape is also in flux, marked by concerns about the exit environment for high-valuation startups and the strategic reallocation of capital by major investors like Jeff Bezos. This dynamic environment highlights a growing distinction between consumer and enterprise AI applications and underscores the critical importance of product-market fit in the AI era.

Navigating Venture Capital in an Unbalanced Market: Insights from a Cyber Security Seed Investor

The venture capital market is currently unbalanced, characterized by escalating entry prices and a high volume of cash that often leads to wasted investments. Despite this, opportunities exist for exceptional growth, particularly in sectors like cybersecurity, where significant pain points drive massive outcomes. Successful early-stage investing prioritizes understanding market dynamics and fostering strong founder chemistry.

Navigating AI Frontier: Leaks, Layoffs, and Liquidity

The AI industry is marked by rapid advancements, strategic pivots, and significant market shifts. Anthropic's accidental leak of its powerful Mythos model highlights the dual challenge of cybersecurity in an AI-driven world and the immense compute demands of advanced AI. Meanwhile, OpenAI's decision to discontinue Sora underscores a strategic recalculation towards more profitable ventures like coding and enterprise AI, against a backdrop of complex internal dynamics and a highly competitive, resource-intensive landscape. These events, coupled with growing concerns over AI's impact on labor markets and the implications of geopolitical tensions on tech investments, paint a complex picture of innovation, risk, and strategic recalibration within the AI ecosystem.

Hims & Hers CEO Andrew Dudum on Disrupting Healthcare with AI and a Long-Term Vision

Andrew Dudum, CEO of Hims & Hers, outlines his strategy for disrupting the healthcare industry by prioritizing direct-to-consumer models, vertical integration, and aggressive market expansion. He emphasizes the importance of building a resilient team, leveraging AI for efficiency, and maintaining a long-term perspective in a rapidly evolving market, challenging traditional healthcare systems and focusing on patient-centric care.

AI Market Dynamics: Anthropic's Efficiency Gain and the 'Big Three' Power Law

Anthropic is rapidly closing the gap with OpenAI, reportedly achieving similar revenue scales in half the time with only 25% of the training costs. This efficiency, combined with OpenAI's management instability and distracted M&A strategy, suggests a shifting competitive advantage toward lean, technically-led AI labs. Simultaneously, the emergence of 'Big Three' titans (SpaceX, OpenAI, Anthropic) is creating an unprecedented power law where their combined IPO values may dwarf all other IPOs from the last two decades.

Effective Intensity Matching for Ionization Mechanism Discrimination

A novel experimental technique, Effective Intensity Matching (EIM) coupled with Intensity Selective Scanning (ISS), enables the comparison of ionization processes in ultrafast laser pulses, independent of pulse ellipticity and diffraction effects. This method allows for direct comparison with theoretical predictions of multiphoton, tunnel, and field ionization, showing high agreement. EIM-ISS also quantifies recollision ionization probability and allows for observation of the influence of precursor ionic states.

Non-recollisional excitation confirmed in multi-electron ionization of krypton

This study provides the first experimental validation of non-recollisional electronic excitation during strong field ionization. Utilizing 50 femtosecond circularly polarized laser pulses on krypton, the researchers observed a high degree of agreement with theoretical predictions. This mechanism is particularly pronounced at laser intensities exceeding 1 PWcm^-2, where the energetic departure of ionized electrons distorts the wavefunctions of bound electrons.

Ultrafast Tunnel Ionization Induces Simultaneous Electron Excitation

This paper presents experimental evidence demonstrating that ultrafast multi-electron tunnel ionization of argon atoms, using circularly-polarized 50 femtosecond laser pulses, results in simultaneous excitation of remaining electrons via a "shake-up" mechanism. This contradicts traditional models that neglect interactions between remaining electrons during ionization. The homogeneous electric field conditions achieved through precise experimental techniques enabled accurate measurement of tunnel ionization probability independent of optical focal geometry, confirming the unexpected electron excitation phenomenon.

Kerr Nonlinearity Induces Optical Non-Reciprocity in Microresonators

This paper details the observation of spontaneous symmetry breaking and non-reciprocal transmission of counter-propagating light in dielectric microresonators. This phenomenon, driven by Kerr-nonlinearity-mediated interactions, results in a resonance frequency splitting that favors one propagation direction. This mechanism offers a fundamental approach to achieving optical non-reciprocity without relying on traditional magneto-optical effects.

Kerr-Effect Symmetry Breaking Enables Magnet-Free Optical Isolation >24 dB in a Single Microresonator

Del Bino et al. demonstrate that the intrinsic nonreciprocity of the Kerr effect in a single ultra-high-Q microresonator is sufficient to build practical optical isolators and circulators — without requiring magnetic materials or external bias fields. Counterpropagating light in a Kerr ring resonator undergoes spontaneous symmetry breaking, forcing circulation in only one direction at a given frequency. The authors achieve over 24 dB of isolation and provide a theoretical power-scaling model for the attainable nonreciprocity, offering a viable path toward fully integrated, CMOS-compatible nonreciprocal photonics.