Jeff Wang
speaker
190 appearances
1 recordings
1 series
first heard Oct 2024
last heard Oct 2024
Jeff Wang’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsNo recordings in the last 12 months.Older appearances are listed below; set an alert to hear about the next one.
Appearances
You are going to have the number two chatbot in the world, you know, before your end in your Instagram, your WhatsApp, your Facebook. You just go right, right now, just go search Meta AI.
I think it's already delivering incremental ARPU. It'll happen first in companies like Meta. Like I said, because you have an auction marketplace, this gets repriced immediately. And so if I look at Meta, we estimate that it's not just delivering revenue. I think it's already delivering about 15 billion of incremental EBIT, EBIT for Meta. Right.
Just in the form of more content recommendation means you have more time on meta properties equals more ad inventory and better ad matching equals higher CPMs and higher CPMs just, you know, it's straight flow through to the bottom line. So it's, it's beautiful already for meta.
Now I look at the software companies and I just think it will take more time because you have to go out to your customer and say, look, I'm delivering you this value. Here's the data. And when your contract comes up again, we're going to raise your prices. It's just a harder conversation.
But what you have seen, even since you recorded the pod with David, Canva, which you mentioned, has raised prices on its enterprise plan by 3x. 3x.
I do agree with it. I think where I share his view is that there is a $600 billion AI problem in the sense that ultimately application companies need to deliver positive ROI from these massive investments. Where I'm probably more optimistic is in the pace that these application companies can actually realize that ROI. We've already talked about Meta. We've already talked about Canva.
I think ServiceNow will start to really flow through some of the ProPlus price increases later this year and into next year as well. I think you're starting to see the initial hints that AI is going to come through in terms of incremental ARPU for a lot of these companies. Now, it's happened probably a little bit more slowly than I would have guessed.
I probably would have been more first half this year as opposed to back half this year. But I definitely think you're starting to see it. And I think you're starting to see it because you see the features that can be productized and the clarity from other companies. What I mean by that, so if you look at what Meta has done is really just take what ByteDance has done. Right.
With a recommendation engine now, like I'm plugging a bunch of GPUs into it. I'm copying the same thing and I'm just rolling out to my customers. Co-pilots, I think you will be able to use what you see at GitHub and be able to roll it out across many different companies.
If you look at enterprise semantic search, if you look at case summarization, these are all features that have been proven to deliver ROI at certain companies. And now you could just pull that into other companies, other products. I think that's going to deliver value.
Now, what I don't know, and this is where I do agree with David, there is a massive amount of capital that's going into not the productization side, but the model training research side. That is where all of the capital is going.
I've been surprised at how the scaling has continued, how just throwing more compute, more data has improved these models. I don't know where that ultimately gets to, but I do think a hundred billion, there are not a lot of companies out there that can spend a hundred billion. What I do believe is you don't need 10 foundation models out there.
I think you need certainly more than one just for the sake of humanity. China will have one. I think the Western world will have at least one. So maybe there's three to five foundation models in the world. It's not going to be 10. It's not going to be 20. And so I do think the...
pool of spend is going to shrink in terms of the number of companies, but the amount that each individual company is going to spend is also going to go up. There is real overinvestment risk, I think.
The way that we've decided to invest in AI in the public markets, we have some investments that are semiconductor or hardware stocks that power AI, but we more so invest in the application companies that over time I think will profitably infuse AI into their products. So if you can buy a nice core business with a call option on AI, I think that is a very good way to play it.
What's the example of that? ServiceNow is a really interesting way to play it. So it's got obviously a very good underlying core business. And I don't know that ServiceNow has necessarily the biggest opportunity in AI, but it's a very clear opportunity. IT ticket deflection is purpose built for AI. To be able to deflect tickets, to resolve them automatically, that is very clear.
That is a very, very clear use case in my mind. I don't know if it will be the biggest use case right now with all the incremental innovation that there is in models. I think coding could obviously be a much bigger use case. But in terms of clarity, ServiceNow's opportunity, I think, is one of the most clear.
Then I'd also argue that overinvestment in the semiconductor and foundation model layer actually should accrue to application companies over time in the form of more better AI capabilities that we can infuse at lower cost.
I think Nvidia's price is reasonable if you think it's gonna continue to keep going. So Nvidia trades at what, 25 times next year earnings. I think the bigger question is what does 26 look like? You know, hyperscaler CapEx is going up. I think it's going up at a rate that is surprising to us. It's grown what, 50% year over year this year. 25 grows another 30%.
Is it 30% higher? I mean, 30% higher, I think would effectively eat away every single dollar of Google EBIT. So what does happen? I think it's a great question. This is where being a hedge fund is actually helpful for investing in something like AI. So we can invest, obviously, as I mentioned, in the application companies.
Showing 121–140 of 190 · page 7 of 10
← Previous
Next →