SoftBank’s $5.8 Billion Nvidia Gamble on OpenAI

According to Forbes, SoftBank has sold its entire $5.8 billion stake in Nvidia to support its $30 billion investment in OpenAI, causing Nvidia’s stock to dip 2%. This marks the second time SoftBank has cashed out of the chipmaker after regretting a 2019 sale that would have netted over $150 billion. Meanwhile, AI customer service startup Sierra has reached a $10 billion valuation and expects $100 million in annualized revenue by January 2026. OpenAI is burning $15 million daily on its Sora video generation tool despite admitting the economics are “completely unsustainable.” The company has committed to $1.4 trillion in data center spending while projecting just $20 billion in annual revenue this year.

SoftBank’s Nvidia history

Here’s the thing about SoftBank’s Nvidia moves – they’ve been here before. They sold a $4 billion stake back in 2019 that Masayoshi Son later called “the fish that got away,” since it would have been worth over $150 billion today. Now they’re doing it again, dumping $5.8 billion to double down on OpenAI. It’s either brilliant timing or spectacularly bad judgment. The fact that they’re also selling $9.17 billion in T-Mobile shares shows how all-in they’re going on this AI bet. But seriously, selling Nvidia to fund AI? That’s like selling Tesla stock to invest in electric cars.

OpenAI’s unsustainable burn

OpenAI is spending money like it’s going out of style. $15 million per day just on Sora video generation? That’s absolutely wild for a company that lost up to $12 billion last quarter. Bill Peebles, who runs Sora, actually admitted the economics are completely unsustainable. But that’s nothing compared to Sam Altman’s $1.4 trillion data center commitment. Let that number sink in – it’s more than the GDP of most countries. According to analysis from Tomasz Tunguz, OpenAI would need to hit $577 billion in revenue by 2029 to justify that spending. That’s Google-level revenue, and OpenAI is currently at maybe $20 billion. Something’s gotta give.

The Altman gamble

What’s really fascinating here is that Sam Altman has nothing to lose financially. He doesn’t own equity in OpenAI, even after their restructuring. As he posted on X, “If we screw up and can’t fix it, we should fail.” That’s easy to say when you’re playing with other people’s money. Experts think the most likely outcome is that OpenAI only uses a fraction of the compute it’s booked, forcing renegotiations with Microsoft, Oracle, and others. These companies don’t want OpenAI to go bankrupt, so they’ll probably cut deals. It’s the classic “if you owe the bank $100 million, you own the bank” scenario. Altman’s playing high-stakes poker, and everyone else is holding the chips.

AI reality check

Meanwhile, the AI gold rush continues. Sierra’s founders became billionaires building customer service bots. Meta’s chief AI scientist Yann LeCun is reportedly leaving because he got bumped down the reporting structure. And sometimes the “AI” isn’t even AI – Fireflies AI admitted their early version was just founders manually taking notes on calls. Basically, we’re in the wild west phase where anything goes and money flows freely. But when you look at the actual numbers – the $15 million daily burns, the trillion-dollar commitments, the regretted stock sales – you have to wonder how much of this is sustainable innovation versus pure speculation. The bubble hasn’t popped yet, but the pressure is definitely building.