Push Your Chips In When You Know It Will Work

After the NV1 flopped in 1995, Nvidia was close to failing, and the RIVA 128 had to work. On the Acquired podcast (October 2023) Huang told the story in his own terms. An emulator company was shutting down “because they had no customers.” Nvidia bought what it had, and “we essentially virtually prototyped the chip by buying this emulator.” (The transcript renders the vendor as “ICOS”; a 2025 book spells it “Ikos,” so treat the name as unsettled.)

The logic. Huang’s account of the decision: “I know it’s going to be perfect, because if it’s not, we’ll be out of business. So let’s make it perfect.” Because the chip had been tested in software before it was manufactured, the company could commit its remaining money to a design it already knew worked.

The result. RIVA 128 shipped in August 1997 and reportedly sold about a million units in four months, a figure from secondary reporting.

The lesson as he tells it. Verification cut the risk. The all-in moment came after the emulator, not before.

Doesn’t transfer as “bet the company.” People remember the boldness. Huang only went all-in after simulating the chip, and it could still have failed for reasons the emulator could not see. Most bet-the-company stories are told by the survivors.