Neil Howe on the SpaceX IPO: Party Like It’s 1999? | Protect the Pile Episode 15
The Protect the Pile team discusses market conditions on SpaceX IPO day, analyzing whether current AI/tech enthusiasm resembles the 1999 dot-com bubble while examining differences in valuations, earnings quality, and public sentiment. They debate the future of AI commoditization, infrastructure bottlenecks, and SpaceX's business fundamentals versus speculative ventures.
Summary
On June 12, 2026, the Hedgei Asset Management team convenes to discuss market dynamics surrounding SpaceX's IPO, broader AI and tech valuations, and historical parallels to the 1999 dot-com bubble. Patrick Kent opens with five key observations: the market is transitioning from quad-two (accelerating growth/inflation) to quad-four (decelerating both), healthcare is showing relative strength, interest rates have peaked and are backing down, energy has returned to bearish trend, and concentration risk in the Magnificent Seven stocks is concerning.
Sam Ramen presents a detailed comparison between 1999 and 2026: while 1999 saw 457 IPOs with 70% average first-day returns and Cisco trading at over 100x forward PE, today's tech giants trade at 20-35x PE with earnings revisions—not multiple expansion—driving gains. Crucially, the number of IPOs is far lower today because private equity and venture capital have allowed companies to have massive private runs before going public at trillion-dollar valuations. Neil Howe contextualizes this through historical railroad bubbles, noting that railroads eventually became commoditized, unprofitable businesses despite transformative impact, and warns that AI—lacking the monopolistic route control railroads had—faces severe price competition risks as OpenAI cuts prices.
The team explores critical differences from 1999: public sentiment is deeply negative on AI (polling below cancer) versus enthusiastic internet optimism then; today's companies are profitable incumbents versus tiny startups; and accounting has become increasingly "tortured" with balance sheets being run through income statements via massive financing. Patrick highlights that hyperscalers must continually raise equity and debt to fund escalating capex commitments, creating a second-derivative growth problem.
On competitive dynamics, the panel debates whether one dominant "god model" AI can exist or if the market will fragment into specialized models. Sam argues that foundry capacity (dominated by TSMC) represents the real bottleneck through 2027-2028, benefiting semiconductor suppliers over hyperscaler spenders. Brooks predicts only three AI companies survive—Google, Anthropic, and OpenAI—as lowest-cost commodity providers, with Meta and others facing existential crises. Neil warns of intellectual property collapse: if AI trains on AI output and steals human creative content without compensation, no one will publish online anymore, killing the web that trains these models.
On SpaceX specifically, Patrick praises Starlink as a genuinely valuable business deserving real valuation but critiques the S1 for inclusion of speculative ventures like lunar mass drivers and orbiting data centers. Neil notes Starlink's strategic vulnerability to EMP attacks and satellite warfare while acknowledging its transformative military and civilian value. Brooks analyzes SpaceX's thoughtful lockup structure for executives, allowing staged selling tied to performance milestones—a model he recommends other companies adopt.
The discussion concludes with acknowledgment that timing these transitions is treacherous: bearish managers in early 1999 were fired by year-end; the best 1999 performers were mostly fired by 2001.
About this episode
In this episode of Protect the Pile, Hedgeye Asset Management portfolio managers R. Patrick Kent, Sam Rahman, Brooks Cutright, and Neil Howe unpack whether today’s AI frenzy looks more like 1999, the railroad bubble, or something entirely new. The group debates market concentration, hyperscaler CapEx, AI commoditization, private-market excess, and how practitioners should navigate momentum without ignoring bubble risk. They also dive into SpaceX’s IPO, Starlink’s real business value, and the far-out promises embedded in the space economy. 00:00 Welcome & episode setup 01:00 Market backdrop: SpaceX IPO day, Iran deal chatter, and risk ranges 02:16 Patrick’s macro notebook: quads, healthcare, rates, energy, and Mag 7 05:37 SpaceX IPO vibes and late-’90s market flashbacks 10:47 Neil Howe on AI as a utility and historical bubble parallels 15:03 Is this 1999? Valuations, IPO mania, and private-market excess 23:04 Why today’s AI bubble is different: incumbents, backlash, and politics 28:02 AI, railroads, monopolies, and the race to build the “God model” 34:16 How practitioners manage AI exposure, momentum, and exit signals 37:36 AI bottlenecks: networking, foundries, hyperscalers, and picks-and-shovels 47:06 IP risk, web content, real-world AI models, and the bullish compute case 52:08 SpaceX IPO deep dive: Starlink, lunar mass drivers, space data centers, and lockups 01:01:05 Final thoughts
Key Insights
- OpenAI's price cuts signal AI is becoming a utility, creating an ominous pattern similar to railroads and high-speed internet where initial transformative technologies eventually become unprofitable due to competition
- Unlike railroads which controlled specific routes, AI competitors face zero switching costs—users can freely move between ChatGPT, Claude, DeepSeek, and others, preventing sustainable moat creation
- Modern earnings are increasingly 'tortured' through accounting with massive balance sheet financing flowing through income statements, making reported profitability less comparable to 1999 startups that had minimal revenue
- Hyperscalers (Google, Meta, Oracle) are locked in an arms race where any company cutting capex risks complete disintermediation within 3-4 years, forcing continued massive spending even if margins collapse
- If large language models train on AI-generated content instead of human-created content, eventual model collapse will occur where AI trained on AI output produces nonsense, and creators will stop publishing online if unpaid
- Taiwan Semiconductor's deliberately constrained foundry capacity expansion creates a structural bottleneck where Nvidia gets priority but Apple, Broadcom, and others face severe delivery delays in 2027-2028
- Public polling shows Americans are more hostile toward AI than any other public globally, despite America generating the AI and being positioned to benefit from revenues—a political vulnerability for companies
- Elon Musk at $1 trillion net worth represents 3.1% of US GDP and 80 basis points of world GDP, comparable to or exceeding Rockefeller and JP Morgan historically, triggering inevitable populist backlash
- SpaceX's S1 filing includes a 'lunar mass driver' definition—an electromagnetic launch system to send materials from the moon to Mars—demonstrating valuation is based on speculative ventures beyond core Starlink business
- Starlink represents a genuinely valuable business that could stand alone at significant valuation, but the trillion-dollar SpaceX price reflects addition of speculative projects like space-based data centers and lunar infrastructure
- In 1999, bearish managers were fired by year-end; the best-performing 1999 managers were mostly fired by 2001, demonstrating that timing entry and exit in bubble cycles is more important than directional correctness
- The real future of AI may involve smaller specialized models proprietary to individual enterprises (JP Morgan's own models, General Electric's own models) rather than one dominant general intelligence owned by a mega-corporation
Topics
Transcript
[0:00] All >> [music] >> right, welcome back to Protect the Pile, the official podcast of Hedgei Asset Management and the only nononsense investment show that cuts through the market noise with a panel of practitioners that play this game every day. We'll dissect the current landscape, debate the opportunities and risks, and help you navigate the markets uh using the hedgei risk framework and lots of real world experience all with [0:30] the goal of protecting and growing your pile. I am Patrick Kent or RPK of hedgei asset management. With me as always is my colleague Sam Ramen. Also on the show today, we have the newest member of the HAM team, Brooks Cart Cutright, as well as…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Hedgeye
The Rebalance Steamroller | Protect the Pile Episode 16
The Hedgei team discusses market dynamics marked by broadening participation beyond mega-cap tech, mechanical rebalancing pressures from Marvell's addition to the S&P 500, and conflicting signals between inflation (declining), yield curve (flattening), and liquidity (tightening). They examine how index mechanics are masking underlying market churn and debate whether conditions resemble shallow Quad 1 or Quad 4 scenarios.
Your Brain Is Always Trying To Hack Your Process
The speaker emphasizes that systematic processes are essential because the brain naturally undermines disciplined trading approaches. Success in investing comes from following simple rules like cutting losses short and letting profits run, while mistakes occur when predictive thinking overrides market signals.
Markets No Longer Align With Fundamental Value
Historical markets aligned with fundamental value because most participants used fair-value models, but passive investment's rise has created a price-insensitive market structure dominated by allocation rules rather than value assessment. This concentration of a single market philosophy creates fragility rather than robustness.