Head of Research @Silicon_Data. @UMICH/@ETH_en/@UVA alum. Perpetually curious, but also "incredibly unsophisticated" (according to Chamath Palihapitiya)
New York, USA
- Tweets 92,216
- Following 5,575
- Followers 59,804
- Likes 167,976
All I saw was “GDM team was working on a new generation of chips to make AI much faster and cheaper”.
I resigned from Google today.
I enjoyed my work and loved the people, but my GDM team was working on a new generation of chips to make AI much faster and cheaper, and I think AI is already progressing too fast, so I had to quit.
1
4
2,423
We had a great conversation with @vikramskr today! Vikram runs both semiexponent.com and semidoped.com, and we’re big fans of his work.
We’re excited to collaborate with Vikram! If you’re also interested in partnering with us on our expert interview database, please get in touch!
Spoke 1:1 with the most excellent @FundaAI today, and I will be doing expert interviews on their platform.
Although most people prefer to keep transcripts to institutional clients, they are letting me post to my Substack, and it will become part of their expert network database!
1
1
20
10,230
Rumor has it that this was the special toy that came with the McDonald’s Happy Meals served at the White House state dinner last night.
💰 Trump's state dinner table seats billionaires worth a combined $1.13 trillion.
Take a look: tmz.com/2026/09/25/trump-sta…
4
23
5,059
Steve Hou retweeted
investmentbrief.ai/p/revenue… - full details here
How much revenue does the AI buildout need in 2030?
Three paths for AI investment
investmentbrief.ai 1
5
2,445
👀
Me+@econJaredB are out w/ a new piece examining how much revenue is needed to justify hyperscaler AI capex. We estimate that incremental AI revenue is currently b/n $86-$188.1B for the hyperscalers, but they need $13.1-$18.7 *trillion* over the next decade for that to pencil out.
1
1
10
6,231
Really enjoyed my conversation with Bill too! Hope you’ll find it interesting too!
I really enjoyed my conversation with @stevehou
Think @Silicon_Data is highly probable to add value to you.
13
4,607
🔥🔥🔥
Tomorrow: @satyanadella
On YouTube, X and @YahooFinance. Daily show launches end of October.
1
21
5,337
This is a very fair point. But I do think it’s worth comparing with how concentrated “cloud computing”, SaaS or even “Paid Ads” remain even so many years after wide deployment. AI has only been around for much shorter period of time.
We live in an inherently and maybe increasingly unequal world with “90-10” all over the place. The percentage of wealth owned by the top 10% vs bottom 90% say. And how many have told us for how long that this was unsustainable?
I think the risk of air-pocket with enterprise adoption of AI is very real. I’ve argued this myself. But it’s also useful to put it within context or “base rate” as @mjmauboussin calls it.
Over the last few weeks, the AI bull discourse has shifted over to AI infrastructure and away from AI models.
Why? AI models are seeing threats to their growth, due to greater competition driving price cuts and model efficiencies. Meanwhile, GPUs and infrastructure are becoming more scarce and rental rates are rising.
I'd argue AI infrastructure is not the safe bet investors think it is. As a spend category, it's even *more* concentrated than AI spend, or any other spend category we track.
The entire bet is based around a narrow group of highly correlated buyers. And simple growth in rental rates or demand does not account for outsized concentration risk.
See for example Ramp data today in Torsten Slok's daily spark
3
6
20
5,087
I do think there's potentially an "air pocket" risk here: as we transition from AI capex being predominantly financed by the two AI labs to being more broadly financed by the opex of hundreds to millions of enterprises. The largest enterprises are the slowest to adopt AI productively due to internal bureaucratic frictions and workflow complexities. The smallest companies will see the fastest AI adoption and the biggest ROI, but they are small. So from an aggregate ROI perspective, the total economic benefits of AI adoption may not get here fast enough to bridge the (perceived) gap of this transition from phase to another.
1
1,605
Zuck seems to be having a lot of fun these days, more so than Sam or Dario it seems.
13
3
136
22,838
We’ve seen a few of this idea come and go: Humane AI Pin, Rabbit R1, and some others I’m forgetting. They all flopped: the AI wasn’t good enough to support the bold functional form so people still find themselves needing to look at the phone. I think this may finally have changed.
Ofc the next thing that’s required is distribution. The previous attempts hardly sold. Meta clearly has distribution but only on software historically and has had a patchwork record of hardware success.
The only company with hardware distribution is ofc Apple. Will Apple get into this lane finally? An Apple-made keychain-sized screen-minimal AI companion device? That could be as big as the AirPods, Watch or even iPhone itself when it finally happens. I think it’s likelier and less intrusive than glasses. Now imagine combining it with AirPods case, which everyone already carries around.
Come to think of it, the AirPods may be a big obstacle to this muse charm thing. Everyone already carries an iPhone and an AirPods case in their pockets. Is there really room for more unless you literally strap it to your wrist or neck as a pendant which ofc makes you look like a crazy person. Tough. We are running out of room on the body.
Things are getting crazy. Meta is releasing Muse Charm, a Tamagotchi-sized 5G device that lets you talk to your AI agent without pulling out your phone.
• Camera + microphones
• 5G connectivity
• Fits on a keychain
• ~2-inch OLED screen
• Fingerprint sensor
Meta says they’ll ship for the holidays in December. No pricing yet.
I’m sure I would want to carry around an extra device when I could just use my phone lol.
26
6
51
15,082