Skip to main content
JL

Jason Lamkin

Harry Stebbings**chinese Open-weight Model Threat**openrouter Sale Timing**founder Sale Decision Framework**infrastructure Vs
7episodes
1podcast

Featured On 1 Podcast

All Appearances

7 episodes

AI Summary

→ WHAT IT COVERS Harry Stebbings, Rory O'Driscoll, and Jason Lemkin analyze five major tech developments: China's Kimi and Qwen open-weight model releases, Washington's response to Chinese AI, Databricks' $3B Series M at $188B valuation, OpenRouter's reported acquisition talks, Fireworks AI's $1.5B round, and Stripe's reported bid to acquire PayPal. → KEY INSIGHTS - **Chinese Open-Weight Model Threat:** Kimi K3 is a 2.8 trillion parameter model comparable in size to US frontier models, not a lightweight alternative. On OpenRouter, 50% of traffic already runs through Chinese-created models. The real risk isn't a sudden takeover — it's gradual enterprise adoption accelerating as cost differentials reach 80% cheaper than closed frontier models like Claude or GPT-4. - **OpenRouter Sale Timing:** The optimal window to sell a routing/aggregation layer business is precisely when commodification begins but before acquirers fully recognize it. OpenRouter's last round valued it near $1.8B; a $5–6B exit represents the crossover moment where hyperscalers — particularly Amazon, which positions itself as model-agnostic — would pay a strategic premium far above standalone NPV to shift enterprise market share. - **Founder Sale Decision Framework:** A 3x return on last round is insufficient justification for a founder to sell. The threshold should be 10x, because the risk-adjusted math changes entirely at scale. A founder holding 12–15% of a $1.8B company walking away with $600–700M at a $5–6B exit faces enormous execution risk for a relatively marginal personal wealth increase versus taking the certain outcome. - **Infrastructure vs. Application Layer Gap:** AI infrastructure spending runs at $800–900B annually. The two frontier model companies generate roughly $100B combined. Every other AI application company combined — including Cursor at $4B ARR — totals under $40–50B. The application layer renaissance has not arrived; spending on training data for foundation models alone likely exceeds the sum of all non-Cursor AI application revenues. - **Foundation Model Growth Rate as Market Bellwether:** The single variable determining the trajectory of US tech valuations, hyperscaler RPO commitments, and AI infrastructure CapEx is OpenAI and Anthropic's revenue growth rate in 2026–2027. If either company slips from 10x annual growth to 2–3x, commitments made against that growth assumption trigger a broad market dislocation across every adjacent sector simultaneously. - **Stripe-PayPal Acquisition Dance:** Stripe's reported offer carries a 28% premium to PayPal's public stock price; historical take-private averages land in the mid-30s. The board's rejection is a procedural negotiating move, not a genuine refusal. Delaware fiduciary duty requires PayPal's board to demonstrate a credible standalone plan exceeding the offer's value — difficult given three consecutive failed CEO turnaround attempts and flat internal metrics. → NOTABLE MOMENT The panel noted that Ben Affleck sold his AI company for over $587M — more than ten times the combined box office earnings from his three highest-grossing films as an actor. The contrast landed mid-discussion about whether to sell a portfolio company for $6B, prompting genuine pause about where value creation actually concentrates. 💼 SPONSORS [{"name": "Base44", "url": "https://base44.com"}, {"name": "Plaud", "url": "https://plaud.ai/20vc"}, {"name": "Fin", "url": "https://fin.ai/20vc"}] 🏷️ Chinese AI Models, Open-Weight LLMs, AI Infrastructure Investment, Stripe PayPal Acquisition, OpenRouter M&A, Venture Capital Valuation

AI Summary

→ WHAT IT COVERS Harry Stebbings, Jason Lemkin, and Rory O'Driscoll analyze Coinbase cutting AI spend 50% while increasing token output, Anthropic's push to ban Chinese open-source models via Senate lobbying, Microsoft's 16% monthly decline, Kalshi's $40B valuation, and Bending Spoons' $20B IPO as a template for B2B SaaS roll-up strategies. → KEY INSIGHTS - **AI Cost Optimization:** Coinbase reduced frontier model spend by 50% in two months by routing workloads to open-source models while maintaining or increasing token output. Every CFO in the Fortune 500 should audit their LLM spend by model tier, implement token routing, and benchmark whether coding-driven spend increases translated into measurable revenue acceleration before approving further budget expansion. - **Frontier Model Revenue Risk:** Anthropic scaled from $1B to $9B ARR in 2024, then to $44B run rate mid-2025, but open-source adoption threatens that trajectory. Companies should evaluate whether their AI spend is concentrated in frontier models for tasks where open-source alternatives perform comparably, since the cost differential can reach 5x, materially compressing margins without proportional capability loss. - **Regulatory Capture Strategy:** Anthropic wrote to the Senate Banking Committee alleging Chinese open-source models distilled their outputs in breach of terms of service, framing it as IP theft and national security risk. The likely policy outcome is a ban on Chinese-origin open-source models for US enterprise use, which would structurally protect frontier model pricing and eliminate the primary low-cost competitive threat. - **B2B SaaS Roll-Up Playbook:** Bending Spoons' consumer roll-up model — buying stagnant assets, raising prices, cutting costs, and installing motivated operators — translates directly to B2B SaaS. Targets like Marketo, PagerDuty, and Asana have sticky customer bases, broken cultures, and flat growth. Buying at 2x revenue, installing a product-focused operator, and adding AI-native features could reaccelerate NRR to justify 8-10x exit multiples. - **Series A Benchmark Reality:** Founders growing from $1.5M to $5M ARR face a structurally difficult Series A environment in 2025. Deals getting swept off the market are growing $1.5M to $15M. At the lower trajectory, founders should expect to pitch 100-150 investors, raise less capital, and consider whether converging toward profitability rather than a growth round better fits their actual curve. - **Claude Tag Enterprise Threat:** Anthropic's Claude Tag embeds an autonomous AI agent directly into Slack channels with access to cross-platform data from Salesforce, HubSpot, and other tools. If the agent captures workflow context continuously and executes autonomously, it could render underlying SaaS applications into passive databases. Enterprises should monitor whether Claude Tag reduces active usage of core SaaS tools within 90 days of deployment. → NOTABLE MOMENT The hosts noted a sharp irony in Anthropic's distillation complaint: the company recently settled litigation with book copyright holders for training on their IP without permission, yet is now lobbying the US Senate to penalize Chinese firms for doing something structurally similar to Anthropic's own models. 💼 SPONSORS [{"name": "Omni", "url": "https://omni.co/20vc"}, {"name": "Checkout.com", "url": "https://checkout.com"}, {"name": "Invisible", "url": "https://invisibletech.ai/20vc"}] 🏷️ AI Spend Optimization, Open-Source Models, Anthropic Regulation, SaaS Roll-Ups, Series A Benchmarks, Prediction Markets

AI Summary

→ WHAT IT COVERS Harry Stebbings, Jason Lemkin, and Rory O'Driscoll analyze DeepSeek's $50B Series A with Chinese government retaining sole voting rights, Wall Street's $725B question about AI ROI, OpenAI's custom Jalapeno inference chip, open source's threat to closed-source models, and why the "flabby middle" of AI pricing creates existential risk for Anthropic and OpenAI. → KEY INSIGHTS - **AI CapEx math:** Hyperscalers spend roughly $700B annually on AI infrastructure, requiring approximately $1.5T in revenue to justify returns. That implies replacing around 8% of the entire US labor force with AI-generated tokens just to break even. By 2027, CIOs will shift from token-maxing experimentation to demanding hard ROI proof before approving further AI budget allocations. - **Open source pricing threat:** Chinese government subsidies effectively fund DeepSeek and five comparable open-source models, making "open source" a misnomer — it's state-sponsored competition. This creates a pricing ceiling on Anthropic and OpenAI's mid-tier offerings. Anthropic is already emailing customers about prompt caching discounts specifically to undercut open-source cost comparisons and retain enterprise workloads. - **Closed-source market structure:** The foundation model market is consolidating into a two-player oligopoly. A third closed-source competitor faces simultaneous pressure from above (OpenAI/Anthropic on quality) and below (Chinese open-source on price). Without a parent company's balance sheet — as Google Cloud has — a standalone number-three closed-source model cannot survive the margin compression. - **LLM moat destruction:** Any competitive advantage built on data lock-in, switching costs, or workflow integration is now vulnerable to LLM-powered migration. Databricks claims a full data lift to their platform in 30 days versus the traditional five-year Accenture-led migration. Founders pitching moats without acknowledging this risk signal a fundamental misunderstanding of the current competitive environment. - **Agentic finance workflow:** A fully functional AI VP of Finance can be built in hours using current models connected to Salesforce, Bill.com, Brex, and QuickBooks — handling quoting, contracting, invoicing, follow-up, payment reconciliation, and bookkeeping without human intervention. Variable contractor costs drop 50–60% incidentally, not through deliberate headcount reduction, as agents absorb tasks humans routinely skip or delay. - **Startup team structure shift:** Winning startups in 2025 run smaller teams at top-of-market compensation with higher equity per person, operating six-plus days per week in-office. The investment calculus has shifted: a company with 40 high-output employees outcompetes one with 100 average performers. Founders still building around remote, part-time work cultures face structural disadvantage against AI-native competitors running continuous sprint cycles. → NOTABLE MOMENT Lemkin described building a fully operational AI finance director remotely from China in a single-digit number of hours — a system that outperformed every human previously handling the role. The agent caught $80,000 in uninvoiced revenue that a human contractor had simply never billed, with no explanation given. 💼 SPONSORS [{"name": "HSBC Innovation Banking", "url": "https://innovationbanking.hsbc"}, {"name": "Deel", "url": "https://deel.com/20vc"}, {"name": "Framer", "url": "https://framer.com/20vc"}] 🏷️ Open Source AI, AI Infrastructure CapEx, Foundation Model Competition, Agentic Workflows, Enterprise AI ROI, DeepSeek

AI Summary

→ WHAT IT COVERS Harry Stebbings, Jason Lemkin, and Rory O'Driscoll analyze Anthropic's $30B raise at a $380B valuation, Thrive's $10B fund close, OpenAI's acquisition of OpenClaw creator Peter Steinberger, Stripe's $140B private valuation versus Adyen's $47B public market cap, and deteriorating public SaaS multiples amid accelerating enterprise AI adoption across Fortune 500 companies. → KEY INSIGHTS - **Anthropic Revenue Trajectory:** Anthropic has achieved three consecutive years of 10x GAAP revenue and run-rate growth, reaching approximately $14B ARR in 2025 from $1B prior — a growth rate with no historical precedent at this scale, surpassing early Microsoft, Google, and Compaq. Investors should note this growth comes without profitability, making the company structurally closer to a semiconductor business than a software company, with compute CapEx commitments running into hundreds of billions over the next three to four years. - **Enterprise AI Adoption as a "Presumption of Success" Cycle:** Fortune 500 companies are committing to AI spend regardless of proven ROI, operating on what the hosts call a "presumption of success." This mirrors hyperscaler behavior two years ago. Investors and operators should expect one to two years of elevated enterprise AI budgets before any retrenchment. The practical implication: companies selling into this wave — particularly AI coding, customer support, and legal tools — face near-zero sales resistance through at least 2026. - **Public SaaS Gravity Well:** The blended annualized growth rate across public SaaS companies has fallen toward 10%, down from 30%+ peaks. At sub-10% growth, these companies enter what the hosts describe as a "dead zone" where valuation support collapses. Operators in horizontal workflow SaaS built pre-2022 should treat current conditions as structural, not cyclical. The practical signal: if your February and March investor updates show no AI-driven acceleration, the venture funding path is effectively closed regardless of existing ARR. - **Stripe vs. Adyen Valuation Framework:** Stripe at $140B versus Adyen at roughly $47B reflects a 2.5x size difference (Stripe ~$5B revenue, Adyen ~$2B), growth premium pricing, and a private market narrative advantage. Adyen runs nearly 50% operating margins with 21% H2 2025 revenue growth. Investors evaluating the pair should note Adyen is measurable and arguably undervalued on free cash flow; Stripe's premium reflects flexibility from staying private and avoiding the public market's simultaneous demand for profitability and AI investment. - **OpenClaw / Autonomous Agent Inflection Point:** Peter Steinberger's OpenClaw demonstrated that removing AI guardrails and enabling semi-autonomous 24/7 desktop agents ignites developer communities even when the underlying technology is replicable in a single day (as Meta's Manus clone proved). The strategic takeaway for founders: the moat is not the guardrail removal itself but the developer ecosystem momentum it creates. Enterprise security vendors building agent-layer controls face immediate demand, as CISOs and chief AI officers now bear direct liability for autonomous agent behavior inside corporate environments. - **Founder CEO Return as AI Transition Signal:** Workday's Anil Bhusri returning as CEO within eight months of departure, alongside UiPath's Daniel Dines, signals that boards view AI product roadmap pivots as founder-specific problems, not generic executive challenges. The pattern: hired CEOs can execute cost and go-to-market playbooks but lack the institutional memory of original architectural trade-offs needed to rebuild core products for AI. Operators at mature SaaS companies should assess whether their current leadership has the specific product knowledge — not just business skills — to execute the transition. - **Monday.com Value vs. Narrative Trap:** Monday.com trades at approximately 10x free cash flow with $1.25B in 2025 revenue, 27% year-over-year growth, and 14% non-GAAP operating margins — statistically cheap for a founder-led, profitable SaaS company. The binary investment thesis: if growth proves durable through AI disruption, the stock is severely underpriced at a 51% year-to-date decline. If enterprise seat expansion stalls as companies shrink headcounts from 6,000 to 2,000 employees, no valuation floor exists. The decision reduces entirely to product roadmap durability, not current financials. → NOTABLE MOMENT Jason Lemkin described an AI sales agent from a portfolio company that, on its first day live, independently identified a target at a major hyperscaler, crafted outreach, and booked a six-figure sponsorship meeting — with zero human involvement. He noted this capability was entirely absent just sixty days prior, underscoring how rapidly autonomous agent performance is compressing what previously required full sales teams. 💼 SPONSORS [{"name": "Dot Tech Domains", "url": "https://get.tech"}, {"name": "Checkout.com", "url": "https://checkout.com"}, {"name": "Invisible", "url": "https://invisibletech.ai/20vc"}] 🏷️ Anthropic Fundraising, Public SaaS Decline, Enterprise AI Adoption, Autonomous AI Agents, Venture Capital Fund Strategy, Founder CEO Returns, Stripe vs Adyen

AI Summary

→ WHAT IT COVERS This episode analyzes major tech deals including Brex's $5.15 billion acquisition by Capital One, Open Evidence's $12 billion valuation, and Anthropic's rising inference costs. The hosts debate whether AI companies can achieve profitability, examine the IPO market's reopening with Equipment Share and Ethos, and discuss implications for SaaS companies competing against well-funded AI-first competitors. → KEY INSIGHTS - **Brex Exit Analysis:** Capital One acquired Brex for $5.15 billion (50% cash, 50% stock), down from its 2021 peak valuation of $12 billion. Despite appearing disappointing versus the 2021 raise, this represents a heroic outcome for founders building to $5 billion before age 30. The deal validates that financial services companies ultimately trade at financial services multiples adjusted for growth, with Brex at approximately 7x revenue on $700 million run rate. - **Hubristic Financing Risk:** Companies raising at peak valuations face a one-day emotional tax when exiting lower, but the alternative of not raising when capital is available would be worse. The strategy works when founders believe they can grow into valuations within two years. Databricks' approach of never raising more than two years ahead of confident valuation targets provides a framework for managing this risk while maintaining competitive positioning against well-funded rivals. - **Ramp Competitive Position:** Ramp's $32 billion valuation faces new scrutiny after Brex sold at 7x revenue. If Ramp maintains $1 billion run rate and faster growth, a 10x multiple at IPO seems reasonable, but Capital One's acquisition of both Discover and Brex creates a formidable competitor with structural cost advantages through closed-loop interchange networks. Ramp must now compete against an A-team with better economics while justifying its 30x+ revenue multiple. - **Inference Cost Reality:** Anthropic's inference costs came in 23% higher than expected, yet gross margins improved from negative 94% last year to positive 40% this year. For B2B companies, inference represents an unavoidable competitive cost that will increase, not decrease, as companies burn more tokens to deliver better agents. Mid-market SaaS companies at $50-200 million ARR face existential challenges funding competitive AI products against rivals with unlimited capital. - **Open Evidence Valuation:** The company raised at $12 billion on approximately $150 million revenue (80x multiple), representing a 12x step-up from its $1 billion valuation earlier in 2025. While the company dominates physician decision support and has clear product-market fit, the direct-to-doctor pharmaceutical advertising market is only $2-3 billion annually. Reaching justifiable public market valuations requires either capturing pharma rep budgets or expanding into adjacent physician services. - **IPO Market Bifurcation:** Equipment Share's successful IPO at $8 billion market cap (growing 47% at $4 billion revenue, profitable) contrasts sharply with Wealthfront's struggling $1.3 billion debut (down 36% from IPO). The market clearly delineates at $3 billion market cap—above this threshold, IPOs proceed smoothly with liquidity; below it, companies face years of illiquidity and talent retention challenges regardless of product quality or mission. → NOTABLE MOMENT One investor revealed shock at seeing which unicorns are actively seeking acquisitions, including companies worth significantly more than their potential acquirers and some with hundreds of millions in revenue showing decent growth. The desperation to exit among 2021-era unicorns has reached levels where founders who appeared confident publicly are privately pursuing any viable exit path, suggesting hundreds of companies remain trapped at unsustainable valuations. 💼 SPONSORS [{"name": "HSBC Innovation Banking", "url": "https://innovationbanking.hsbc"}, {"name": "Deal", "url": "https://deal.com/20vc"}, {"name": "Framer", "url": "https://framer.com/20vc"}] 🏷️ M&A Valuations, AI Infrastructure Costs, IPO Market, SaaS Competition, Venture Capital, Financial Services Tech

AI Summary

→ WHAT IT COVERS NVIDIA acquires Groq for $20 billion, Meta buys Manus for $2.5 billion, OpenAI spends 46% of revenue on stock compensation, Navan trades at 4x ARR, and invisible unemployment emerges as AI reshapes labor markets in 2026. → KEY INSIGHTS - **Strategic Semiconductor Acquisition:** NVIDIA paid $20 billion for Groq despite only $175 million in revenue because eliminating potential margin pressure is worth under 20% of annual free cash flow. The deal closed in two weeks before Christmas at exactly 3x the last funding round to remove objections instantly and secure low-latency inference capabilities before competitors could respond. - **AI Orchestration Valuation:** Manus sold to Meta for $2.5 billion at 25x ARR with founders owning 80% equity, choosing local maximum over risk. Founders recognized orchestration layers face competition from Anthropic and OpenAI, making half a billion dollars each with zero capital gains tax in Singapore versus uncertain future growth against well-funded competitors building similar capabilities. - **Compensation Without Ownership:** OpenAI spends $1.5 million per employee on stock compensation, 34x higher than comparable pre-IPO tech companies, because CEO Sam Altman owns zero shares and prioritizes winning over dilution concerns. This enables aggressive talent retention against $20-50 million offers from Meta, though 60% of researchers still leave within the first year despite no vesting cliffs. - **Private Market Premium Persists:** Revolut generates $3.5 billion in annual profit at $75 billion private valuation while comparable public company Chime trades at $6 billion, demonstrating private markets still offer cheaper capital than public markets. Founders can dividend out hundreds of millions annually without selling shares, eliminating incentive to endure public market scrutiny and quarterly reporting requirements. - **Invisible Unemployment Emerges:** Entry-level jobs disappear as companies like Shopify achieve growth three consecutive years without adding headcount, while senior executives with 2021 toolkits quietly exit the workforce. Stanford computer science graduates without AI training struggle to find employment while top 0.1% talent receives infinite offers, creating visible tension among highly educated, articulate 23-24 year olds facing unprecedented job scarcity. → NOTABLE MOMENT One investor's AI assistant spontaneously named itself Ren and now searches 14 months of conversation history to provide answers, leading to the realization that most knowledge workers will run AI inference 24 hours daily by year-end, fundamentally transforming how people work and justifying massive infrastructure investments. 💼 SPONSORS [{"name": ".tech domains", "url": "https://get.tech"}, {"name": "Checkout.com", "url": "https://checkout.com"}, {"name": "Invisible", "url": "https://invisibletech.ai/20vc"}] 🏷️ AI Acquisitions, Venture Capital Exits, Labor Market Disruption, IPO Markets, Semiconductor Industry

AI Summary

→ WHAT IT COVERS Databricks reaches $100B valuation at 25x revenue while growing 50% annually. Discussion covers CoreWeave's $11B debt financing, Nubank's $2.5B profit milestone, OpenAI's $6B staff secondary, and whether AI infrastructure spending can sustain trillion-dollar projections. → KEY INSIGHTS - **Private Market Valuations:** Databricks at $100B growing 50% with $4B ARR trades at 25x revenue versus Snowflake's 26% growth at similar scale. The valuation appears reasonable if growth persists at 40-50% for two to three more years, reaching normalized multiples on $10-12B revenue base. - **AI Tool Consolidation Risk:** Companies now deploy 10+ AI agents costing $500K-$1M annually at $60-100K per tool. Budget fatigue will drive rapid consolidation within 12-24 months, faster than SaaS consolidation cycles, favoring platforms like Rippling offering multiple agents with orchestration layers over point solutions. - **CoreWeave Debt Structure:** The $11B debt raise requires matching long-term customer contracts with debt duration to avoid classic banking mismatch risk. Success depends on Microsoft and OpenAI honoring seven-year take-or-pay commitments. Any quarterly weakness signals broader AI infrastructure demand slowdown before hyperscalers show strain. - **Labor Replacement Pacing:** AI adoption at forward-leaning companies shows real human replacement, but enterprise adoption requires 18-24 month cycles. Products must address large enough headcount pools to justify CFO attention—high ROI on small workstation counts fails despite efficiency gains. Surface area of automation matters more than percentage efficiency. - **Fintech Geographic Dynamics:** Nubank reaches $60B market cap serving 123M customers with full banking services in weak LatAm incumbent markets. Revolut targets FX/crypto in moderately efficient Europe. Chime reaches only $11B focusing on deposits in well-run US banking market, demonstrating outcome size correlates directly with incumbent weakness. → NOTABLE MOMENT One participant reveals their small team now employs 10 AI production agents replacing five humans at $500K annual cost, with only one human attending daily standups. This shift from theoretical discussion to practical implementation demonstrates how quickly AI labor replacement materializes at forward-leaning organizations. 💼 SPONSORS [{"name": "Qualified (Piper AI SDR)", "url": "https://qualified.com/20vc"}, {"name": "HubSpot", "url": "https://hubspot.com/ai"}] 🏷️ AI Infrastructure, Fintech Valuation, Enterprise AI Adoption, Venture Capital Strategy, SaaS Consolidation

Explore More

Frequently Asked Questions

What podcasts has Jason Lamkin appeared on?

Jason Lamkin has appeared on 1 podcast we summarize, including 20VC (20 Minute VC) — 7 episodes in total. Every appearance is listed below with an AI-generated summary.

Does Jason Lamkin appear as a guest speaker on podcasts?

Yes. Jason Lamkin has been a guest on 1 show we track, across 7 episodes. Browse each appearance below to read the key takeaways and listen to the original.

Where can I find summaries of Jason Lamkin's interviews?

Read AI-generated summaries of all 7 of Jason Lamkin's podcast appearances on SignalCast — each with key insights and a link to the full episode.

Never miss Jason Lamkin's insights

Subscribe to get AI-powered summaries of Jason Lamkin's podcast appearances delivered to your inbox weekly.

Start Free Today

No credit card required • Free tier available