20VC with Harry Stebbings
Showing 1–15 of 711 transcripts.
- 1h 9m
Google Loses $500B in 12 Hours | DoorDash Buys Instacart for $17B | Kalshi Raises $1B
Jason Lemkin, Rory O'Driscoll, Harry Stebbings
Recent market corrections have shifted the AI narrative from euphoria to tempered expectations as OpenAI and broader sector valuations adjusted while pricing pressures intensified for Anthropic. Amidst these volatility shifts, major corporate maneuvers included DoorDash's consolidation of Instacart and Salesforce's aggressive acquisition strategy to modernize legacy stacks against rising agentic disruption risks. Simultaneously, investment committees like Andreessen Horowitz are recalibrating portfolios to favor high-revenue velocity over speculative valuations, signaling a pivot toward sustainable fundamentals in both software and emerging legal or robotics sectors.
- 1h 2m
Modal CEO: Has Open-Source LLM Inference Peaked? | What Happens to the Cost of Compute
Erik Bernhardsson, Harry Stebbings
Modal Financial has achieved $1 billion in annual recurring revenue and secured a $15.75 billion Series D valuation driven by its proprietary infrastructure and $300 million run rate from its Sandbox product line. Co-founder Eric Vissio projects the inference market will expand to a $100 billion opportunity while the company aggressively scales GPU capacity and targets 95th-percentile compensation to compete in a talent-intensive landscape. Looking forward, Modal aims to diversify beyond inference into training and reinforcement learning services, betting on a fragmented market where speed and developer experience outweigh traditional software moats.
- 1h 21m
David Tisch: Why No Manager Wants to Stay Small & Why Competing with Mega Platforms is a Losing Game
David Tisch, Harry Stebbings, Jason Lemkin, Rory O'Driscoll
David Tish of Box Group redefines the current venture landscape as a unique, innovation-driven cycle where concentrated seed funds must surrender ownership targets to 1–2% to compete against multi-stage giants offering larger capital. His firm employs a distinct "founder-friendly" strategy that prioritizes long-term alignment over pro-rata rights, operating within a market consolidating toward a top tier of 20–30 firms capable of accessing essential AI infrastructure. This approach has yielded outsized returns, exemplified by the $8 billion exit of Cursor, while challenging outdated methodologies like the Lean Startup in favor of substantial capital deployment for technical founders.
- 1h 0m
Instinct Worth 30BN? | Microsoft Cuts 27,000 Jobs While Meta Taps Equity Markets for $95B CapEx
Jason Lemkin, Rory O'Driscoll, Harry Stebbings
Anthropic's filing reveals a $16.3 billion first-half revenue driven by rapid model cycles and massive compute expenditures, establishing a financial barrier that outpaces competitors like OpenAI while facing market dynamics disrupted by Chinese data breaches. Meanwhile, the sector undergoes significant structural shifts as Meta executes a $95 billion equity raise to fund infrastructure, Microsoft offers voluntary buyouts to pivot its workforce, and hyperscalers prepare for future capital depreciation. Investment committees debate valuations ranging from a $1.25 trillion OpenAI bet to a vetoed Instinct AI round, signaling a divergence between established AI giants and high-risk ventures in an increasingly capital-constrained market.
- 24 min
Satya Nadella: Whether to Trust Frontier Models | Why Companies Need to Own Their Own Intelligence
Satya Nadella, Harry Stebbings
Satya Nadella outlines a strategic shift where AI alignment is treated as a distinct engineering challenge requiring "engineered zero trust" architectures with immutable logs and embedded safety in model weights. He predicts a market evolution toward licensed weights and private evaluation benchmarks, driven by the commoditization of token usage and the need for firms to capture value from their own data exhaust. Consequently, the global landscape will likely see horizontal specialization across the stack, with immediate supply bottlenecks in silicon and power while sectors like software development lead productivity gains that could lift global GDP.
- 1h 13m
Inference Will Be $10TRN Market: What No One Sees | Software Investors are F***ed | Naveen Rao
Naveen Rao warns that the AI industry faces an imminent energy crisis driven by grid constraints and soaring infrastructure costs, forcing a strategic shift from hardware commoditization to energy-efficient co-design and specialized silicon. He argues that while software margins are collapsing, future value will concentrate on energy conversion capabilities, robotics, and verification layers within a $10 trillion inference market. Rao concludes that deep tech investment must prioritize technical progress over traditional validation to build the physical infrastructure required for a post-transformer era where AI accelerates scientific discovery rather than posing existential threats.
- 1h 0m
The New VC Game | CoreWeave Sell-Off Creates Panic | Revenue Multiples Out of Control
Jason Lemkin, Rory O'Driscoll, Harry Stebbings
Current venture capital structures have narrowed investment criteria to billion-dollar positions with trillion-dollar exit potential, marginalizing traditional vertical SaaS in favor of ultra-high-risk infrastructure plays. While companies like CoreWeave face valuation pressures due to debt and demand concerns, major players such as Anthropic and Modal demonstrate robust growth metrics that justify revised multi-billion dollar valuations despite market volatility. Strategic shifts by figures like Elon Musk, alongside public market dynamics, are reshaping investor expectations for long-term returns and defining new benchmarks for private asset multiples.
- 1h 20m
Who Wins in the Next Decade of VC | Margin and Price Matter Less in a World of AI | Sameer Dholakia
Sameer Dholakia, Harry Stebbings
Bessemer Venture Partners recently closed a $5.75 billion fund structured to capitalize on Samir's forecast of a trillion-dollar AI supercycle, asserting that massive capital deployment is the primary competitive moat in an increasingly consolidated sector. The firm is aggressively deploying capital into high-conviction investments like Anthropic, Agora, and Waymo while anticipating a structural shift where companies remain private until public market demand for pure-play AI exposure peaks between 2026 and 2028. By prioritizing hyper-growth multiples and large single-position allocations, Bessemer aims to navigate a landscape where intelligence will eventually exceed oil as the world's largest market.
- 1h 3m
The Future of Inference Market & Model Routing: Commodity or Increasing in Value? | LangChain CEO
Harrison Chase, Harry Stebbings
This session analyzes Langchain's operational strategy, detailing how engineers manage the 30% salary cost of coding tools through visibility, budget caps, and harness engineering to prevent costly misuse scenarios like a recent $40,000 incident. Speakers outline a technical architecture relying on neutral harnesses and hierarchical model routing while identifying vertical agents and memory layers as the primary drivers for future investment and infrastructure. The discussion concludes with founder insights on securing durable brand monopolies through execution speed and predictions that multi-agent interactions will eventually drive exponential increases in total inference demand despite falling per-unit costs.
- 1h 17m
Valar Atomics Sues Day One Ventures | Meta's Back to the AI Frontier | NVIDIA's $30B Reflection Deal
Jason Lemkin, Rory O'Driscoll, Harry Stebbings
Valor Atomics initiated a contentious legal dispute with seed investor Day One Ventures over pro-rata rights waivers in its $6 billion Series B, exposing strategic friction between large megafunds and early backers. Simultaneously, the broader tech landscape faces consolidation as NVIDIA pursues open-source acquisitions to lock in ecosystem spend, while Meta's internal "Muse Wonder" model reinforces a leadership dynamic centered on Mark Zuckerberg and Alex Wang. Investors are adjusting to a deflationary AI market driven by Anthropic's price cuts and rising IPO revenue thresholds, prompting capital shifts toward high-utility robotics and drug discovery firms like Isomorphic Labs.
- 59 min
Atlassian CEO: Enterprises Won't Trust Chinese AI | Why The Jury’s Still Out on Specialised Models
Mike Cannon-Brookes, Harry Stebbings
Atlassian CEO Mike Cannon-Brooks rejects the "headless" AI strategy, arguing that design and user experience must remain critical differentiators as the company builds a foundational interface layer and top-tier Model Context Protocol server for its "Teamwork Graph." Instead of replacing humans, Atlassian positions its agents as context enablers that integrate across tools like Figma and Databricks while prioritizing accountability through rigorous management of agent provenance and cost-effective model routing. Looking ahead, the company anticipates a shift from information asymmetry to imagination asymmetry, where enterprises leverage semantic search of live organizational data to foster high-value creative work amidst rapidly declining inference costs.
- 1h 12m
How to Build a Sales Machine | Max Freeman, Ramp
Ramp redefines sales operations by treating talent acquisition and pipeline management as engineering problems, utilizing rigorous vetting simulations and a high-performance compensation model that pays SDRs double the market rate to book 40 to 60 meetings monthly. The company implements a strict performance tiering system where underperformers are immediately terminated, while new hires undergo intensive 60-to-90-day bootcamps to master proprietary automation tools before accessing accounts. By delaying verticalization until capturing significant market share and enforcing a dual-threat leadership model, Ramp achieves forecasting accuracy within a 5% delta while compressing sales cycles through aggressive disqualification of prospects lacking executive commitment.
- 1h 16m
Cognition vs Factory | Anthropic Under Threat | ElevenLabs Doubles Its Valuation to $22B
Dev Ittycheria, Jason Lemkin, Rory O'Driscoll, Harry Stebbings
OpenAI's accelerating revenue growth contrasts with a delayed 2027 IPO, while Anthropic targets a November 2024 public offering and Salesforce executes a $2 billion acquisition of Listen Labs to capitalize on rapid AI application exits. Concurrent market turbulence includes the withdrawal of Aura's IPO filing and the erosion of executive loyalty norms, exemplified by a board observer's swift transition from Factory to competitor Cognition. Additionally, the sector faces emerging legal challenges regarding Grok's acquisition structure and shifting consumer behaviors driven by the Jevons Paradox, which fuels increased AI usage despite falling costs.
- 1h 11m
Can You Still Win in Venture Without a $1BN Fund? Menlo’s Venky Ganesan on the New Rules of VC
Venky Ganesan, Harry Stebbings
Menlo Ventures employs an aggressive "buy everything" strategy to participate in defining AI companies, prioritizing the ability to scale positions in outliers over maintaining high initial ownership stakes. Partner Venky Ramachandran guides this approach by emphasizing data-backed conviction, rapid capital deployment, and the selection of founders capable of communicating complex concepts simply. The firm accepts dilution and lower equity percentages at the seed stage to secure entry into massive opportunities like Anthropic, aiming to maximize returns through asymmetric upside rather than early-stage control.
- 1h 5m
Crusoe CEO: Why Everyone Gets GPU Depreciation & AI Energy Costs Wrong
Chase Lochmiller, Harry Stebbings
Crusoe Energy secured a $3.9 billion Series F round to reach a $30.9 billion valuation while pivoting from Bitcoin mining to a vertically integrated AI infrastructure model. Founder Chase Lockmiller leads the company's strategy to bypass traditional supply chain delays by manufacturing power distribution internally, enabling the rapid deployment of a 1-gigawatt-scale computing facility in Abilene that promises to deliver capacity significantly faster than the industry standard. This approach addresses critical constraints in skilled labor and physical connectivity through modular construction and closed-loop cooling, aiming to generate substantial local tax revenue while managing GPU asset lifecycles through diverse revenue streams.