Deep Dive
SpaceX's sudden collapse and the AI capital question
Ed Elson opens with SpaceX's stunning 30% drop in just days, wiping out $400 billion in a single day—the second worst selloff in stock market history. The stock has recovered slightly to sit 22% below its highs, but the damage signals deeper concerns. Gil Luria, head of tech research at DA Davidson, explains the root cause: SpaceX just issued a $25 billion bond offering to finance AI operations after already raising $85 billion in equity. This aggressive capital raise is triggering investor anxiety about the fundamental question haunting the entire AI boom—where are the returns? If companies need to raise this much money just to build compute capacity, can they actually monetize it and justify these valuations? The timing matters: this drawdown is happening before any lockup expirations, meaning the real selling pressure from insiders hasn't even begun.
The broader tech rout and market confusion
The tech selloff extends far beyond SpaceX. The NASDAQ 100 dropped roughly 4% for the week, shedding over a trillion dollars in two days. Semiconductor stocks got hammered hardest—AMD down 5%, Nvidia down 5.5%, TSMC and Broadcom down 8%—with the pain spreading globally to Asian indices. Luria characterizes this as massive volatility driven by extreme dispersion in outcomes for 2026. If AI works, US GDP could grow 5%; if the cycle stalls, growth drops to 1-2%. Usually investors debate whether GDP hits 2.8% or 3.2%—now the range is catastrophic. But here's where it gets weird: big tech like Microsoft, Amazon, and Google all sold off despite telling investors their AI data center returns are already locked in with known markups. Investors simply aren't buying the narrative that spending on chips will pay off.
Google's talent drain and the AI leadership question
A critical moment emerges around Google's sudden selloff. Two senior AI researchers left Google for Anthropic and OpenAI, prompting investor panic. Luria makes the crucial point: there are probably fewer than 100 people globally who can develop frontier AI models, and the two who just left Google are likely in the top 10. When they publicly stated Google is too bureaucratic and DeepMind won't lead to superintelligence, they signaled serious cracks in the narrative that made Google the AI winner. That narrative had powered Google's stock from $180 to over $350. This exodus isn't just talent drain—it's a credibility hit to the entire moat Google investors believed it had constructed. The broader market implication: if even Google can't retain top AI talent, how confident should investors be in any big tech company's AI timeline?
Market inconsistency: AI is both worthless and priceless
Luria cuts to the market's central contradiction. Nvidia, Microsoft, Micron, and Amazon's valuations imply AI won't work and the cycle is peaking in 2024—classic peak-cycle pricing. Meanwhile, pure-play AI stocks like Cerebras and optical companies are valued as if the cycle runs unchecked through 2030. The market is being internally inconsistent. Skeptics won't buy the big companies; believers only want the marginal players. Luria sees this as an opportunity: companies like Microsoft and Nvidia are trading at genuinely attractive valuations if AI actually works out. But the messaging is muddled. SpaceX trades at 100x revenue (absurd), while Microsoft trades at historically low multiples (attractive). How do you make sense of a market where the most capital-intensive AI play crashes on debt issuance while the software giants that claim they already have monetized returns also slide? Until the market resolves whether AI delivers or collapses, volatility will remain extreme.
Minimum wage data reveals real-world evidence
The episode pivots to Arindrajit Dube, economics professor at UMass Amherst, who studied a natural experiment: 30 states raised minimum wage (averaging $14-plus per hour) while 20 stuck with the federal $7.25 floor since 2007. The standard economic argument says higher wages kill jobs. Dube's 2010-2025 data tells a different story. Restaurant pay in the 30 raised states is now 8-9% higher than in the 20 frozen states, a clear wage growth differential. Yet restaurant jobs per capita—the most sensitive sector—remained virtually flat from 2013 to 2025 in both groups. Same employment, higher wages. Neighboring-county comparisons across state lines show identical results: cross a border, wages spike up, employment doesn't budge. Dube explains employers absorb the cost through three mechanisms: productivity gains offset some costs, profits compress, and prices rise modestly. A burger costs a bit more, but overall cost of living barely moves. The evidence is airtight for this natural experiment.
Why minimum wage hasn't killed jobs and what comes next
Dube's research demolishes the main argument against raising the minimum wage. But he's careful: this is just one natural experiment, and he's done more granular work comparing low-wage industries and broader wage baselines. The story doesn't change. He points to the three Ps: productivity offsets some costs, profits take a hit, prices adjust slightly. Most people support higher minimum wage once they understand these tradeoffs. The bigger implication: if the federal minimum hasn't budged in 17 years while inflation eroded purchasing power, that's a policy failure, not a market decision. Dube advocates raising the federal minimum and, critically, indexing it to inflation so Congress doesn't have to rehash this debate every generation. Full employment also matters—workers with alternatives have leverage to demand raises. The political will is the blocker, not the evidence. Multiple states continue raising minimums through ballot measures, but not all states allow this, making federal action essential.
SpaceX lockups and the Nasdaq inclusion catalyst
Elson returns to SpaceX in the final segment with a reality check on his pre-IPO prediction. Part one nailed it: the stock popped 30% on day one, hitting $176, and kept rising to $220 the following week. Part two predicted a 50% crash over six months due to valuation nonsense and lockup expirations. The stock is already down 30% and still hasn't faced the real test: when thousands of early investors can finally sell and unlock billions in liquidity. At that point, the question becomes simple—hold an irrationally valued stock or sell and buy a house, boat, or plane? Elson knows the answer. One bright spot: SpaceX will join the Nasdaq 100 in two weeks, triggering roughly $8 billion in automatic purchases from passive index funds. That temporary demand boost might provide a brief bid. But after that catalyst, SpaceX faces the greatest selling pressure in stock market history as insiders unload shares. The bond offering already spooked investors; imagine their nervousness when the real selling begins.