20VC: SpaceX Launches Largest Ever IPO | OpenAI Files to Go Public | Uber Cuts 23% of HR | Lovable Hits $500M ARR | Founders Revolt Against VCs: The Fundraising Horror Stories Going Viral
Harry Stebbings, Jason Lemkin, and Rory O'Driscoll discuss the week's biggest tech news including SpaceX's $1.77 trillion IPO roadshow, OpenAI filing to go public, Uber cutting 23% of HR, Lovable hitting $500M ARR, and founders sharing VC fundraising horror stories. The panel debates IPO mechanics, AI efficiency trends, and the structural shift toward leaner startups powered by AI.
Summary
The episode opens with analysis of SpaceX's historic $75 billion IPO roadshow at a $1.77 trillion valuation. The panel highlights an unusual mechanic: Elon Musk pre-fixed the share price at $135 rather than using traditional price discovery, which Rory argues significantly raises the probability of the IPO trading flat or down on day one. With the book reportedly only 2x covered — far below the typical 8-10x — the panel is cautious about short-term performance, though all three acknowledge SpaceX as a generationally significant company. Jason speculates it won't pop dramatically in week one, while Rory predicts the stock is unlikely to hold its IPO price 12 months out given it's priced at roughly 70x forward sales.
On OpenAI's confidential S1 filing, the panel interprets the 'flexible timeline' framing as expectation management rather than genuine hesitation. Rory suggests the internal mandate is almost certainly to move as fast as possible. The broader point raised is that all major AI companies are hitting a scale where public capital markets become necessary, and the window of favorable market conditions is being seized simultaneously.
The discussion of Uber cutting 23% of HR and reinstating a three-day in-office policy leads to a broader conversation about AI-driven workforce efficiency. Jason argues HR is ripe for AI disruption, while Rory notes the more significant Uber story is its continued push into autonomous robo-taxi deployments in Europe. The panel connects this to the broader theme of companies using AI to shrink headcount while maintaining or growing output.
Lovable hitting $500M ARR with only 146 employees and Cursor targeting $6B ARR become the centerpiece of a debate about the new economics of AI-native startups. Jason argues that any founder or board member who justifies needing 200+ more headcount deserves scrutiny, while Rory adds the nuance that high token costs replace labor costs in these businesses, making direct revenue-per-head comparisons to traditional SaaS misleading. Both agree that enterprise-focused companies will inevitably need larger go-to-market teams, but expect the baseline efficiency to be 2-5x better than prior generations.
On the founders-vs-VCs fundraising horror stories, Jason empathizes with the emotional depth of founder rejection but argues that fundraising is fundamentally a sales process and founders must develop thicker skin. The Cloudflare CEO's story about Vinod Khosla's team suggestion is discussed — with Jason noting the comment, while poorly timed during a pitch, reflects a common VC behavior of reading team dynamics, and the outcome proves it was incorrect. Rory notes that turning down 99% of deals makes high customer satisfaction structurally impossible for VCs.
Other topics covered include: Ramp raising at $44B valuation with over $1B ARR and strong growth; Suno raising at $5.4B despite Jason finding the valuation hard to justify from a utility standpoint; Bending Spoons filing for a US IPO at $20B — described as a 'consumer Vista' that acquires legacy brands like AOL, Evernote, and Vimeo, cuts costs, and aggressively raises prices; Databricks raising at $165B while staying private because its capital needs are orders of magnitude smaller than foundation model companies; and Microsoft launching its own AI models, which the panel views as a necessary but belated step, with questions remaining about whether Microsoft can grind its way to 'good enough' as it did with Azure.
Key Insights
- Rory argues that SpaceX's pre-fixed IPO price removes the traditional price discovery mechanism that bankers use to engineer a first-day pop, meaningfully raising the probability of flat or negative opening day trading.
- Jason claims that if the SpaceX IPO book is only 2x covered, it signals insufficient demand at the stated price, and he does not expect a dramatic pop in the first week, though he believes retail enthusiasm may drive long-term appreciation.
- Rory predicts SpaceX stock is unlikely to hold its IPO price 12 months out, citing a base rate where IPOs priced above 10x forward sales frequently decline, and SpaceX being priced at approximately 70x forward sales.
- Jason argues that OpenAI's 'flexible timeline' framing around its S1 filing is primarily expectation management for the press, while internally the mandate is almost certainly to complete the IPO as quickly as possible.
- Rory contends that Elon Musk turned a potential loss into a win in Q1 2025 by converting excess data center capacity — originally built without a competitive foundation model — into approximately $24 billion annually in compute revenue from Anthropic and Google.
- Jason argues that any startup founder or executive who justifies needing significantly more headcount to scale deserves scrutiny, pointing to Lovable's 146-person team at $500M ARR as evidence that the efficiency bar has fundamentally shifted.
- Rory notes that high token costs in AI-native businesses structurally replace labor costs, meaning revenue-per-head comparisons between companies like Lovable and traditional SaaS like Salesforce are misleading without accounting for the different cost structure.
- Jason claims that the majority of Anthropic's enterprise sales do not require human sales interaction, suggesting that even enterprise-focused AI companies can maintain leaner go-to-market teams than traditional software companies.
- Rory describes Bending Spoons' strategy as a consumer version of the Vista/Tombravo private equity playbook: acquire legacy consumer software brands, cut marketing and overhead, raise prices aggressively, and rely on high-inertia customers who are unlikely to churn.
- Jason argues that founders who were treated poorly during fundraising should reframe the experience as a sales process — noting that losing a sales deal is normal — while reserving genuine grievance only for cases where VCs caused harm post-investment, such as forcing out founders.
- Rory argues that because VCs turn down 99% of deals they see, high customer satisfaction is structurally impossible for the asset class, and rejection being the default MO means some percentage of poor interactions is statistically inevitable.
- Rory states that the imperative for foundation model companies like OpenAI and Anthropic to access public capital markets is categorically different from software companies like Databricks, because the capital requirements are two to three orders of magnitude larger, making private funding increasingly insufficient.
Topics
Full transcript available for MurmurCast members
Sign Up to Access