Tesla is targeting production capacity of 1 million robots per year, capacity that could match roughly half its entire 2025 vehicle output, but for an unproven product. A slower ramp doesn’t just delay revenue, it stretches the burn on a parallel manufacturing buildout (Fremont plus a second, larger factory at Giga Texas targeting summer 2027, with long-term capacity of 10 million units annually) while that capacity sits underutilized. Slower ramp = capex committed earlier than revenue arrives = worse near-term free cash flow. This is structurally similar to the AI hyperscaler capex-vs-payback tension

Technicals support the data

Not bad. I have sell orders at 258. 

it correctly reports Burry’s five short positions but omits that his stated thesis is a valuation call (semiconductor index 65% above its 200-day average, Caterpillar’s 30-year-high price-to-sales ratio) rather than a bet that AI itself fails. It also doesn’t mention Burry’s history of headline-sized bets turning out smaller in practice (e.g., his 2021 Tesla puts were often overstated as “hundreds of millions” when actual capital was just the option premium), which matters for weighing how large this “wager” really is.

CoreWeave's biggest buyer became its competitor. Jane st & NVDA have been backing an unprofitable venture

Meta’s move comes weeks after SpaceX made a similar move via xAI, and the pattern suggests the winners of the AI race may be the ones who own the data centers, rather than the ones with the best models, but only if compute demand continues to hold and data centers retain their value. This entire Meta story remains an unconfirmed report