Another day, another pile of AI headlines, and today's batch is all about money and trust. Nvidia's sniffing around a huge check for Perplexity, SoftBank's taking on debt to cover its OpenAI bet, and Meta's finally admitting its smart glasses had a creep problem. Let's get into it.
Nvidia is circling a $30 billion-plus stake in Perplexity
Word is Nvidia is in talks to put equity into Perplexity at a valuation north of $30 billion. That's a big number for a company that started the year with less than $250 million in annualized revenue and is now reportedly running above $750 million. Tripling revenue in under eight months is no small feat, especially in a search and answer-engine market that Google could squash any given Tuesday if it wanted to.
Why it matters: Nvidia keeps finding ways to stay tangled up in every corner of the AI boom, not just chips. Owning a slice of the companies buying your hardware is a heck of a hedge. If those companies win, you win twice.
Robert's take: I like Perplexity as a product, I use it myself some days over Google. But a $30 billion valuation on $750 million in revenue is priced for perfection, and perfection is a hard thing to keep delivering when you're competing against Google, OpenAI, and now every browser maker slapping AI search into their product. Nvidia writing this check tells you more about how much cash Nvidia is sitting on than it does about Perplexity's moat.
SoftBank is selling a record bond to cover its OpenAI bet
SoftBank is planning a 1 trillion yen retail bond, about $6.3 billion, to pay off the bridge loan it took out to fund its OpenAI stake and to keep making more AI deals. That's the largest retail bond of its kind, sold straight to everyday Japanese investors.
Why it matters: SoftBank has gone all in on AI being the next trillion dollar platform, and Masayoshi Son does not do things halfway. Financing an OpenAI stake with debt, then refinancing that debt with more debt sold to retail investors, is a pretty clear sign of how much leverage is stacked under the AI boom right now.
Robert's take: This is the part of the AI story that makes me a little nervous. When the money backing the biggest AI bets starts looking like a game of financial musical chairs, somebody eventually doesn't have a seat. SoftBank tried something similar with WeWork, and that didn't end great. OpenAI is a stronger business than WeWork ever was, but leverage is leverage, and retail bondholders are the ones left holding the bag if the music stops.
Meta is patching its AI glasses so they can't be used to secretly film people
Meta announced its Ray-Ban style AI glasses will now stop recording the moment someone covers the capture LED, the little light that's supposed to tell people nearby they're being filmed. The update is aimed squarely at covert filming and harassment complaints.
Why it matters: Smart glasses with a built in camera were always going to run into this problem. People figured out fast that you can tape over or dim that LED and record folks without them knowing, which is exactly the nightmare scenario privacy advocates warned about when these things launched.
Robert's take: Good on Meta for actually fixing this instead of just saying the light is there so it's not their fault. But this should have been solved before the glasses shipped, not after the creepy filming stories started piling up. Wearable cameras aimed at strangers need guardrails built in from day one. Bolting on trust after people already got burned is the AI industry's oldest habit, and it needs to stop being the norm.
That's the roundup for today. Big money chasing bigger valuations, bigger debt chasing that money, and hardware makers playing catch up on privacy. Same story, different day, but worth watching close.