E91: SoftBank's $21B+ Vision Fund loss, signals of a bubble, macro picture, Trump raided by FBI

SoftBank’s Vision Fund had just lost $21 billion in one quarter, while Masayoshi Son said six trillion yen had disappeared across two quarters. Son admitted he became “somewhat delirious” during the boom, as capital deployment plunged from roughly $20 billion to $600 million year-over-year.

Chamath argued the fund’s deepest structural mistake was forcing a $100 billion vehicle through roughly five years of new investments. That implied about $20 billion annually, ballooning check sizes and compressing diligence; he argued a 10-year investment period would have allowed far greater selectivity.

SoftBank often gave startups far more capital than they sought, turning money itself into a company’s main competitive weapon. Freeberg argued this pushed acquisition costs upward, damaged unit economics and encouraged businesses such as WeWork to expand beyond what their underlying models could economically absorb.

Saudi Arabia and Abu Dhabi were described as unusually protected Vision Fund investors because much of their capital sat in preferred equity paying roughly a 6% coupon. Chamath’s point was that those investors could earn substantial returns even if the underlying fund merely broke even, leaving SoftBank with more downside exposure.

The clearest bubble signal they identified was “valuation trophy hunting”: founders and investors celebrating markups and unicorn status instead of business quality. Sacks added that median public SaaS valuations rose from roughly 6× forward revenue historically to 15× in 2021, while high-growth companies moved from about 8× to 35×.

Freeberg saw rising consumer debt, rather than the recession label itself, as the macro risk worth watching. Household debt exceeded $16 trillion as variable-rate borrowing became costlier; he argued defaults could become dangerous if rising debt service began outpacing household income and asset gains.

Chamath said he began selling in November 2021 and believed that decision “saved my ass” over the following six to eight months. He resisted confident macro forecasts, arguing that interacting risks across energy, housing, rates, credit and inflation made precise predictions unreliable.

On the Mar-a-Lago search, the panel’s shared concern centered on institutional trust rather than a settled judgment about Trump’s conduct. Sacks repeatedly said he lacked enough evidence to judge whether the search was warranted, while Freeberg focused on how distrust of federal institutions could deepen polarization regardless of the eventual facts.