WRONG: All-In Podcast Just made BIG Mistakes on Iran & AI.

TL;DR

Meet Kevin fact-checks the All-In podcast's March 13 episode, exposing factual errors on Iran oil prices, AI revenue figures, and basic math.

Key Points

• 1.Chamath falsely credited Trump's comments for oil falling from $120 to $90/barrel, ignoring that IEA releasing 400 million barrels from global strategic reserves drove the drop.
• 2.David Sacks claimed the entire Arabian Peninsula relies on desalination, ignoring Yemen — roughly one-third of the region's ~93 million population — which cannot afford desalination.
• 3.Brad Gerstner claimed Anthropic had a $6 billion month in February (implying $72B ARR), but latest reports show Anthropic's ARR is $14–19 billion, making Brad's figure 3.8–5x inflated.
• 4.Chamath said AWS has '12 nines of accuracy,' but Amazon only advertises 11 nines for data durability (not uptime or AI accuracy); actual AWS server uptime yields ~8.7 hours of downtime per year.
• 5.Chamath claimed sanctioned Venezuelan and Iranian oil covers 100% of China's transportation sector, but transportation uses only ~50% of China's oil, and Russia supplies nearly half of China's total oil needs.
• 6.David Sacks incorrectly said Vitalik Buterin donated Dogecoin to charity; he actually donated Shiba Inu (SHIB) — $100 million worth to the Future Life Institute and India's COVID relief fund in 2021.
• 7.David Sacks made a basic math error claiming a 5%/year wealth tax over 20 years leaves you with zero (5×20=100%); the correct calculation is 0.95^20, leaving 35.8% of wealth, still a 64.2% loss.
• 8.Chamath's claim that one gigawatt of AI data center capacity in Arizona costs $50 billion is exaggerated; Cushman & Wakefield estimates standard data centers cost ~$11.7B/GW, roughly $20B with Nvidia chips.