Most important take away

A Wall Street Journal report says OpenAI is missing internal revenue targets even after a fresh $122B raise and an aggressive IPO push, raising serious questions about whether companies tied to its spending (Oracle, CoreWeave) can ever get paid. Meanwhile General Motors quietly delivered a strong quarter with a growing high-margin software business (Super Cruise, OnStar) and trades at roughly 6.4x forward earnings, making it a more grounded investable idea than the AI ecosystem trades.

Summary

OpenAI / AI ecosystem

• A WSJ report indicates OpenAI is missing aggressive internal user-revenue projections. Management has previously guided to ~$280B of revenue by 2030 — more than NVIDIA generates today — and the hosts view that as unrealistic.
• OpenAI just raised $122B yet is already moving toward an IPO, suggesting capital needs are far larger than recent raises have covered.
• Knock-on risk: companies whose stories depend on OpenAI spending — Oracle (ORCL) and CoreWeave (CRWV) — are under pressure. Oracle is already down ~50% from its September pop on the OpenAI deal announcement, reflecting investor skepticism that OpenAI can pay for what it has contractually committed to.
• DeepSeek’s updated open-source model reportedly matches/exceeds OpenAI and Anthropic on some benchmarks at roughly one-fifth the compute cost. Hosts take the exact number with a grain of salt but agree the conversation is shifting from raw capability to cost efficiency. Sora getting de-emphasized at OpenAI is cited as evidence cost discipline is starting to matter.
• Lou’s framework: for current AI valuations to work, one (or some combination) of three things must happen — (1) models become dramatically less compute-hungry, (2) hyperscalers gain real pricing power despite competing with each other, or (3) current valuations are simply sustainable. Matt: if you’re a creditor in this ecosystem, be nervous.
• Actionable insight: be skeptical of names whose thesis rests on OpenAI’s contractual revenue (Oracle, CoreWeave). First-mover advantage in AI may not hold — Netscape/AskJeeves/Yahoo were dominant before Google; OpenAI could be displaced similarly.

General Motors (GM) earnings

• Adjusted EPS $2.82 vs $3.35 prior year, but beat expectations after a ~$500M expected tariff refund and ~$1B one-time EV-strategy pivot charge.
• Margins strong despite a tough consumer environment; GM’s buyer incentives are at the low end of the industry (less discounting than peers).
• #1 US market share overall; clear #2 in EVs behind Tesla — Cadillac EV sales up 20% YoY, GM EV market share now 13% (up from 10% sequentially).
• Raised full-year guidance to ~$12.50 EPS at the midpoint, putting GM at ~6.4x forward earnings.
• High-margin software is the underappreciated story: Super Cruise paid subscriptions up 70% YoY, targeting 850,000 by year-end; OnStar has 13M paid subscribers. Lou’s caution: auto features historically migrate from premium to standard, which could compress those subscription margins over time.
• International weakness — GM China sales -22% YoY. BYD earnings -55%, illustrating how brutal the Chinese EV market is. Hosts see a future where US automakers dominate the US but struggle abroad.
• Actionable insight: GM (NYSE: GM) flagged as a value name — strong execution, raised guidance, low multiple, optional upside from software/services. Watch sustainability of Super Cruise/OnStar pricing power.

Stocks/tickers mentioned

• OpenAI (private, IPO pending) — skeptical
• Oracle (ORCL) — skeptical, down ~50% since Sept on OpenAI-deal feasibility doubts
• CoreWeave (CRWV) — caught in the same OpenAI exposure
• NVIDIA (NVDA) — referenced as revenue benchmark
• Anthropic — throttling users due to compute cost
• Alphabet (GOOG/GOOGL) — Matt prefers non-voting class C shares (cheaper, same economics); class B shares give insiders 10x voting power, a governance flag
• General Motors (GM) — bullish; ~6.4x forward earnings, raised guidance
• Ford (F) — mentioned alongside GM as US pickup/SUV strength
• Tesla (TSLA) — EV market leader
• BYD — earnings -55%, cautionary read on Chinese EV competition

Mailbag — Proxy voting

• Question from listener Jett: how should individual investors think about proxy voting?
• Matt Frankel: doesn’t weight voting rights heavily; will buy non-voting share classes if economics are equal; does care about dual-class structures (e.g., Alphabet) as a governance input to thesis.
• Lou Whiteman: admits he rarely votes — clunky brokerage UX, ~80 holdings — but agrees you should.
• Tyler Crowe (soapbox): if you own individual stocks, read the proxy and vote — board, exec comp especially. Voting “no” on egregious comp packages is empowering. Cites Benjamin Graham’s chapter on shareholder duty.
• Actionable insight: at minimum read your proxies; consider voting against excessive executive compensation; factor dual-class voting structures into your investment thesis.

Chapter Summaries

1. OpenAI struggles & AI ecosystem risk — WSJ report on missed targets, $122B raise, IPO push, knock-on hits to Oracle and CoreWeave, DeepSeek’s cost-efficient model, and the three scenarios needed to justify current AI spending.
2. General Motors Q1 earnings — beat on adjusted EPS, tariff refund/EV charge adjustments, strong margins, low incentives, #1 US share, raised guidance to ~$12.50 EPS, Super Cruise/OnStar growth, China weakness, BYD comparison.
3. Mailbag — proxy voting — three takes on whether individual investors should vote their shares; consensus that governance and exec comp votes matter even if individual votes rarely swing outcomes.