20VC: Will Jensen Return to NVIDIA | Anthropic Set for Largest IPO of All Time | Kalshi, Mercor & Fractile: Mega-Rounds Continue
20VC: Will Jensen Return to NVIDIA | Anthropic Set for Largest IPO of All Time | Kalshi, Mercor & Fractile: Mega-Rounds Continue
Podcast1 hr 12 min
Listen to Episode
Note: AI-generated summary based on third-party content. Not financial advice. Read more.
Quick Insights
  • NVDA: Stay cautious until there is more clarity on leadership and U.S. export controls; monitor China exposure and whether AI investment and acquisition plans continue.
  • Anthropic (private): Treat the anticipated $1.4–$1.5 trillion IPO valuation as a high-expectations, high-risk opportunity; assess final pricing, compute costs, and growth before investing.
  • HUBS, MNDY, and CRM: Watch new-customer growth closely; the discussion points to pressure on software demand, and a founder return alone may not reverse HubSpot’s slowdown.
  • PLTR: It was cited as a public-software growth standout, but the discussion offered no valuation analysis or price target, so do not treat that comparison alone as a buy signal.
  • Across AI investments, favor companies showing measurable customer demand and durable revenue, and avoid assuming private-company valuations or anticipated acquisitions guarantee returns.
Detailed Analysis

NVIDIA (NVDA)

  • The discussion was uncertain about Jensen Huang’s departure and a possible return. One guest said Huang would be their preferred CEO, but doubted the board would reverse its decision quickly.
  • The main business risks raised were U.S. export-control scrutiny and potential loss of access to China-related revenue. One guest estimated China could represent roughly 15% of revenue, while another suggested the broader China-related exposure could be larger; the transcript did not resolve the figure.
  • Guests argued that NVIDIA’s aggressive investments and deals helped support the AI ecosystem. They said Huang’s departure could weaken that activity and make some expected startup acquisitions or backstops less certain.
  • The stock reportedly lost as much as $601 billion in market value in one morning after the news.

Takeaways

  • The discussion points to leadership, export controls, and China exposure as key issues to monitor. A return by Huang was presented as possible, not assured, and was not expected by all guests to restore the company’s prior position.
  • For investors, the transcript offers a cautious, risk-focused view rather than a direct buy or sell recommendation. It also raises a broader question: whether NVIDIA’s past role as a buyer and backer of AI companies will continue.

Anthropic (Private; IPO discussed)

  • The guests described an anticipated IPO at roughly $1.4–$1.5 trillion, with the offering reportedly 7–10 times oversubscribed. These were reported figures and expectations in the discussion, not confirmed final IPO terms.
  • One guest said Anthropic could reach a $100 billion revenue run rate and compared its projected workforce of fewer than 6,000 employees with Google and Meta, which had substantially more employees at similar revenue levels. The discussion used this as evidence of AI’s potential to change company productivity and employment.
  • The company was described as a potential winner from weaker competition in open-weight AI and from constraints on certain ways of accessing advanced chips. The guests also cited a reported $10 billion training cost for a recent model and said about 75% of compute went toward developing the next model.
  • The IPO was seen as a major signal for venture capital: it could encourage limited partners to redeploy capital and investors to pursue larger, faster-growing AI companies.

Takeaways

  • Anthropic was discussed with strongly bullish sentiment, but the proposed valuation and reported demand imply substantial expectations. Investors should distinguish the company’s growth story from the uncertainty around IPO pricing, compute costs, and future model spending.
  • The episode’s broader point is that AI companies may scale with fewer employees, but the guests disagreed on how far that trend will extend across the economy.

SpaceX (Private)

  • Guests praised Starship as a major technical achievement, but one said the rocket business represented less than 10% of SpaceX’s market capitalization.
  • They argued that a SpaceX investment case increasingly depends on compute infrastructure and demand, rather than rockets alone. One guest framed the key question as whether Anthropic would keep buying more compute or cancel it.
  • Starlink was described as differentiated, particularly in less crowded areas, but the discussion emphasized compute as the largest part of the company’s market opportunity.

Takeaways

  • The episode’s view was interested but valuation-conscious: technical progress alone was not considered enough to justify buying at the discussed price.
  • The key diligence question raised was whether expected compute demand will materialize and persist. The transcript gives no specific SpaceX price target.

Kalshi (Private)

  • Kalshi was discussed at a $65 billion valuation and described by one guest as a market leader in prediction markets, with sports betting as a major opportunity.
  • Supporters argued that the U.S. sports betting market is large and that states could become reliant on the associated revenue, potentially making a future ban politically difficult.
  • A skeptical guest highlighted regulatory risk and lawsuits, and questioned how large the addressable market could be even if prediction markets were fully legalized.

Takeaways

  • The discussion was mixed: strong growth and market potential weighed against material regulatory uncertainty and a high valuation.
  • Investors considering the opportunity would need to assess how much of the company’s future depends on legal and regulatory outcomes, rather than treating the overall betting market as automatically available to Kalshi.

Mercor (Private)

  • Mercor was discussed at a $20 billion valuation. Guests viewed it as a strong business in the data-services market, with proximity to AI labs helping its position.
  • One guest said AI companies could each spend around $1 billion or more on training data, creating a large opportunity for data providers.
  • The counterpoint was that Mercor may depend on a small number of large customers and that it was difficult to judge the company’s value at the proposed price. The discussion also described the work as “dirty work” that AI labs may not want to do themselves.

Takeaways

  • Sentiment was positive about the business, but less certain about valuation. The case depends on sustained AI training-data spending and Mercor maintaining its position with major labs.
  • The guests favored investing at the discussed price, while acknowledging that their conviction partly reflected the company’s recent success rather than certainty about future returns.

Fractile (Private)

  • Fractile was discussed in connection with a $720 million financing at a $7.5 billion valuation. The company is developing inference chips to compete with providers such as NVIDIA and Cerebras.
  • A guest cited a reported $250 million order from Anthropic as evidence of demand. Supporters argued that specialized chips could gain share as AI inference expands.
  • A more cautious guest questioned the company’s technology edge and noted that it had no revenue at the time discussed. They compared it with Cerebras, which they said had already shipped chips and was generating substantial revenue.

Takeaways

  • The opportunity was presented as high-potential but high-risk, not as a value investment. The key questions are whether Fractile’s chips deliver a durable performance advantage and whether the reported order turns into repeatable revenue.
  • The transcript’s comparison with an established competitor underscores the importance of validating Fractile’s technology and commercial progress against shipped products and current revenue.

Cerebras (Private)

  • Cerebras was presented as a more established inference-chip comparison to Fractile. One guest cited a valuation in the mid-$20 billions, a strong recent quarter, and projected revenue of just under $1 billion for the year, potentially rising to $3 billion the following year.
  • The discussion acknowledged uncertainty: one guest said they did not know enough about Fractile’s technology edge to decide which company offered better value.

Takeaways

  • Cerebras was framed as a more commercially proven comparator, though the revenue figures and forecast were guest-cited, not guaranteed outcomes.
  • The transcript suggests investors compare chip startups on both technical differentiation and evidence of deployment—not just the size of the AI-chip market.

Hugging Face (Private)

  • NVIDIA’s potential acquisition of Hugging Face was discussed as part of Jensen Huang’s effort to support open-weight AI. Guests said the deal might not close.
  • They identified regulatory uncertainty as a concern, particularly if NVIDIA becomes entangled in disputes over chip exports. One guest also questioned whether NVIDIA’s current leadership would be as motivated to pursue the acquisition.
  • The discussion noted that regulatory scrutiny could delay or complicate a deal, even if the strategic rationale remains.

Takeaways

  • The potential transaction was viewed as uncertain, with regulatory and leadership risks affecting the likelihood of completion.
  • The broader lesson for private-company investors is that an expected strategic acquisition should not be treated as a guaranteed exit or downside protection.

HubSpot (HUBS)

  • HubSpot was described as under pressure after a sharp drop in quarterly net-new customers: from roughly 10,000 per quarter in prior years to 6,000, then 5,000, with guidance below 5,000.
  • Guests said the return of co-founder Brian Halligan could restore founder-level urgency and help the company reposition for AI. They also cautioned that it would be difficult to reverse a slowdown at a large, established software company.
  • One guest argued that small businesses may be delaying or avoiding traditional software purchases, rather than replacing every product by building their own software.

Takeaways

  • The discussion was cautious. A founder return may help, but the metric to watch is whether HubSpot can stabilize and improve customer acquisition—not just raise prices or reposition its product.
  • The episode also raises the possibility that some small businesses may defer purchases of conventional software when lower-cost alternatives seem adequate.

Salesforce (CRM)

  • Guests said Salesforce has been effective at selling multiple products to existing customers, but questioned its ability to attract new customer accounts; they noted that the company had stopped disclosing its customer count.
  • One guest described building custom interfaces on top of Salesforce with AI tools and suggested that some businesses may use agents to complete tasks that previously required marketing-automation software.

Takeaways

  • The discussion points to pressure on traditional software purchasing, especially if AI agents let customers get by with existing systems or delay buying additional tools.
  • Investors may want to watch customer growth and evidence that AI features generate new demand, rather than assuming established customer relationships alone will protect growth.

Monday.com (MNDY)

  • Monday.com was cited alongside HubSpot as experiencing a steep decline in net-new customer additions. One guest cited a fall in a particular customer segment from roughly 1,200 to fewer than 200; the exact segment was described as “10-plus seat” customers.
  • The guests debated whether this reflects customers building their own tools. One view was that the more immediate issue is simpler: small businesses may just be choosing not to buy certain software products.

Takeaways

  • The episode’s tone was cautious, with customer acquisition the main signal to monitor.
  • A slowdown in new customers could reflect changing buying priorities, not necessarily direct replacement by AI-built software.

Microsoft (MSFT)

  • Microsoft’s Outlook, Teams, and Copilot services were said to have been down for roughly 36 hours. Guests described the cause as a network or configuration issue, rather than an AI-related cyberattack.
  • The outage prompted discussion about the reliability of mission-critical technology and the need for stronger operations, security, and infrastructure.

Takeaways

  • The episode did not offer a direct view on Microsoft’s valuation or stock. Its investment-relevant theme was that reliability and resilience may require greater spending as software becomes more central to business operations.
  • The outage was presented as a reminder that operational failures can affect even large technology companies.

ServiceNow (NOW)

  • A possible 40% layoff at ServiceNow was mentioned, but the guest said they were not sure whether the report was confirmed.
  • The discussion used the report as an example of how AI may change staffing needs, while emphasizing that the claim was still uncertain.

Takeaways

  • Treat the reported layoffs as unverified, not as a confirmed company development or investment signal.
  • The broader discussion suggested monitoring how AI-related efficiency changes affect software companies’ costs, staffing, and demand.

Palantir (PLTR)

  • Palantir was cited as an exception among public software companies, with one guest saying it was doubling while other established SaaS companies struggled to reaccelerate above 30% growth.

Takeaways

  • The company served as a positive comparison point, not a detailed investment recommendation. The transcript did not provide a price target or valuation analysis.

Venture Capital and AI Investment Theme

  • Guests described a market with fewer deals but much larger total dollars: they cited Q3 2026 data showing deal count down 21% while dollars invested rose 2.5 times.
  • They said limited partners may increasingly judge venture funds by how many $20 billion-plus companies they hold, encouraging concentrated bets on a small number of perceived winners.
  • Several guests warned that rapid fundraising and fast deployment could drive up startup valuations and compress returns, even if companies continue to grow quickly.
  • The NVIDIA discussion also raised concern that strategic acquisitions and other expected exits may be less dependable. One guest said this could weaken the practical value of liquidation preferences—the contractual priority some investors receive in a sale or liquidation.

Takeaways

  • The episode favored the importance of finding large AI winners, but also highlighted concentration, valuation inflation, and exit uncertainty as risks for venture investors.
  • Investors should not assume that a high-profile strategic buyer will acquire or rescue a startup, or that preferred investment terms guarantee a favorable outcome.

AI Agents and Enterprise Software Theme

  • Guests described a shift toward AI agents that can carry out business workflows, potentially reducing the need for some standalone software purchases.
  • They said businesses may prioritize tools that directly automate labor-intensive tasks, such as customer support or collections, over software they view as less essential.
  • The discussion also highlighted security and reliability risks as agents become more capable and connected to company systems. Guests worried that smaller vendors or poorly secured integrations could create vulnerabilities.

Takeaways

  • The investment opportunity discussed is in AI products that deliver measurable business outcomes, not simply products labeled as AI.
  • The risks raised include security breaches, service outages, high compute costs, and difficulty retaining customers once an agent becomes embedded in workflows.

Other Venture and Acquisition Mentions

  • Revolut and Zipline were praised as attractive deals from the prior week, but the transcript did not provide valuations or detailed investment terms.
  • OpenAI’s acquisition of Edesia, a travel-focused company founded by Sonder’s founder, was discussed as an example of acquiring specialized capabilities for AI agents. No deal price was given.
  • Meta was presented as a formidable competitor in AI agents because of its distribution and ability to fund product development. The guests did not provide a valuation view or a direct stock recommendation.

Takeaways

  • These examples reinforce the episode’s focus on AI-related strategic acquisitions, agent products, and venture-backed companies, but the transcript does not provide enough financial detail to assess the specific deals.
  • No cryptocurrency investment thesis or specific coin recommendation was discussed.
Ask about this postAnswers are grounded in this post's content.
Episode Description
AGENDA:  00:00 Nvidia's $600B Wipeout: Pressure Mounts for Jensen's Return 14:00 Anthropic's $1.4T IPO: A New Record for Venture Capital 19:00 The $20B Club: Unicorns Are No Longer Enough 23:00 Kalshi, Mercor & Fractile: Mega-Rounds Divide Investors 29:00 Nvidia's Deal Spree Under Threat: AI Startups Lose Their Safety Net 32:00 HubSpot's Founders Return as Customer Growth Slumps 41:00 Small Businesses Shun SaaS: AI Agents Win the Budget 45:00 Microsoft Outage & FAA Shutdown Expose Fragile Infrastructure 54:00 OpenAI Buys Odessia: The End of Vertical Personal Assistants? 56:00 Meta's Muse Takes Aim at the $100K Software Stack 01:06:00 SpaceX Catches Starship: The Stock Is Now a Bet on Compute
About The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch

The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch

By Harry Stebbings

The Twenty Minute VC (20VC) interviews the world's greatest venture capitalists with prior guests including Sequoia's Doug Leone and Benchmark's Bill Gurley. Once per week, 20VC Host, Harry Stebbings is also joined by one of the great founders of our time with prior founder episodes from Spotify's Daniel Ek, Linkedin's Reid Hoffman, and Snowflake's Frank Slootman. If you would like to see more of The Twenty Minute VC (20VC), head to www.20vc.com for more information on the podcast, show notes, resources and more.