Sam Altman: ‘It’s Hard To Imagine the AI Revolution Without Jensen’ | Silicon Valley & The World — Transcript
Full transcript
- 0:00Um, great room. Sam, thank you for
- 0:02coming.
- 0:02>> Thank you for having me.
- 0:03>> Um, so just curious, did you catch
- 0:05Saturday Night Live this weekend?
- 0:06>> I just did.
- 0:08What did you think?
- 0:09>> Uh, it like a little funny. I love
- 0:13Saturday Night Live.
- 0:14>> A little funny. Okay. Well, so that was
- 0:17about the Jeden Wong uh scandal. Um, you
- 0:20a few years ago, almost exactly three
- 0:22years ago, uh, had a situation where you
- 0:24were also ousted by the board. Uh there
- 0:27was a letter going around about Jensen
- 0:29that was I think copy pasted from the
- 0:30one that your supporters drafted a few
- 0:33years ago. Um so I'm curious, do you
- 0:36think he should do you think Jensen
- 0:38should come back?
- 0:40>> Um
- 0:42I don't think I know enough to answer
- 0:44that question. I Jensen has long been a
- 0:47hero of mine. Heroes have flaws. I am
- 0:50grateful for the company he's built. Um
- 0:53I
- 0:55I do have like personal strong empathy
- 0:58to a situation like this. I remember uh
- 1:02what it was like although it feels like
- 1:03forever ago. Um I
- 1:06but it's hard to imagine the AI
- 1:08revolution without Jensen. Um and the
- 1:12company he has built took an early bet
- 1:15on us. Uh we took an early bet on him. I
- 1:17have extreme extreme gratitude for that.
- 1:21They also guaranteed I think $ 105
- 1:24billion for your data center in Ohio.
- 1:27They're also, you know, they're the
- 1:28bank, they're sort of backing the entire
- 1:30industry. Do you think that guarantee is
- 1:32worth as much today as it was, you know,
- 1:35a couple weeks ago.
- 1:37Um,
- 1:39I mean, Nvidia is like a behemoth of a
- 1:42company. I think people still trust
- 1:44their guarantees very much. But but
- 1:45yeah, I mean Nvidia without Jensen like
- 1:47Jensen and Nvidia clearly very deeply
- 1:49associated uh and
- 1:54I I think the ambition he has brought
- 1:56and the scale has has been quite
- 1:58important to the industry.
- 1:59>> So you think there there will be a
- 2:01ripple effect there'll be an impact on
- 2:03the industry. Um,
- 2:05look at this point I think there's so
- 2:07much demand in and it's so clear that
- 2:11like AI is real and people want to use
- 2:13it and the revenues and demand are
- 2:14growing so strongly uh that I think the
- 2:18industry will be fine uh but again I
- 2:21think Jensen has played this
- 2:24extremely special and valuable role as
- 2:27someone to really drive things forward
- 2:28with great ambition. So anthropic going
- 2:32public this week. Um in a time when I
- 2:34think there's there's a lot of
- 2:35skepticism from investors. Um curious,
- 2:39do you actually want to go public? Like
- 2:42if you could just avoid it, would you?
- 2:45>> I think there's a lot of skepticism from
- 2:46journalists. Um investors seem to be
- 2:49pretty bullish on AI. I uh things will
- 2:53go up and down. Um prices will go up and
- 2:55down. You know, there will be periods of
- 2:57overbuild and underbuild. Um there will
- 3:00be periods where something goes wrong or
- 3:03something goes really right. But the the
- 3:05overall trend is that huge value is
- 3:09being created. Um huge value is being
- 3:12disrupted too. And I think investors
- 3:15seem I just as we were backstage
- 3:17listening to the amount of capital being
- 3:19invested. Investors seem to realize
- 3:21that. Um do I want to go public? We we
- 3:24will go public at some point. I
- 3:27I think there are real downsides to
- 3:29being public. Uh
- 3:33and it's a very volatile time in our
- 3:35industry with very high stakes
- 3:36consequential decisions to make around
- 3:40safety and how society's going to
- 3:41interact with this technology. Um and
- 3:43there are doubtless things that are
- 3:44easier as a private company. On the
- 3:47other hand, uh I think it's not just
- 3:49like investors that h happen to have
- 3:51access to private markets that want to
- 3:54invest. the world as a whole would like
- 3:56exposure to this and I think they're
- 3:57that's you know directionally a good
- 3:59thing to do.
- 3:59>> Yeah. Do you think if so Anthropic is
- 4:02going to go first? Uh that's clear. Once
- 4:05you go public, is there enough, you
- 4:07know, do you think there's a just enough
- 4:09investor demand left or are you going to
- 4:11be sort of the lift to their Uber, do
- 4:12you think? Is that a worry?
- 4:14>> That is not a worry.
- 4:17And again, I I get that you have to like
- 4:18write something every day and there's
- 4:20got to be like, oh, this, you know, this
- 4:22fight and that one and is this winner
- 4:24and this loser and what's happening here
- 4:25and there, but like the macro trend and
- 4:28the value being created. Um, I am not I
- 4:32am not at all worried about there not
- 4:33being a huge amount. In fact, you know,
- 4:36one way you could look at the last year
- 4:37is there's been way too much investor
- 4:40demand and not enough supply. Um but you
- 4:43know I I think we'll be able to continue
- 4:44to build we the whole industry just
- 4:47incredible incredible value.
- 4:49>> Yeah. Do you So you made these these
- 4:51comments behind closed doors it was
- 4:53reported to Samsung executives where you
- 4:55said there's been some double counting
- 4:57in the industry. What did you mean by
- 4:59that? Can you explain?
- 5:00>> Sure. uh when
- 5:04for a long time uh there was basically
- 5:07one chip provider in the industry and uh
- 5:11so when people were counting wafers they
- 5:14could ask companies like ours what
- 5:16demand was going to be um and they could
- 5:18I think get a sense of where end user
- 5:21demand for the chips they could make
- 5:23would be um there are now many people
- 5:25making chips and many of those people
- 5:28think they're going to sell most of the
- 5:30volume to us anthropic and others. Um,
- 5:33and we have, you know, some limited
- 5:35amount of demand and data centers and
- 5:38everything else. And I think there are
- 5:39many people making chips all talking to
- 5:42the silicon supply chain. Uh, assuming
- 5:44that we're going to buy all of their
- 5:46chips. And, uh, I think the demand, the
- 5:51data center buildout, that stuff is
- 5:53solid. There are,
- 5:55but we're not we're not all going to be
- 5:57able to or need to buy every, you know,
- 5:59all of our supply from every chip
- 6:00vendor.
- 6:01>> What does that mean for the industry?
- 6:02Are people is it over, you know, are
- 6:05things overpriced? What do you think?
- 6:07>> Um,
- 6:09again, I'm sure there will be ups and
- 6:10downs. The long-term demand for huge
- 6:14amounts of compute. Um, and and the
- 6:17value that that unlocks, just the
- 6:18incredible stuff that people are now
- 6:20doing at at real scale with this
- 6:22technology, I think that will keep
- 6:24going. Um and you know like any other
- 6:29like any other tech revolution um the
- 6:31long-term trend is clear the forecast
- 6:33any quarter to quarter and year to year
- 6:36of supply and demand exactly matching.
- 6:37Harder to say it's interesting you say
- 6:39you know they could count on us buying
- 6:41all the chips. I thought you did want to
- 6:42buy all the chips. Um but do you think
- 6:45like one one way of looking at this
- 6:46whole Nvidia thing is now you can.
- 6:49They're not shipping to China anymore.
- 6:51that seems to be
- 6:51>> um
- 6:52>> the supply chain is very complicated.
- 6:55We've been constrained by different
- 6:56things at different times. Sometimes
- 6:58it's been memory, sometimes it's been
- 6:59rack assembly, sometimes it's been, you
- 7:01know, data center capacity. Um sometimes
- 7:06it's been some random component like
- 7:08you're like really we can't make you
- 7:10know we can't sell more services because
- 7:12the world is out of like PCB laminate or
- 7:14whatever. Um, so it's it's just a very
- 7:17complex supply chain um at now at huge
- 7:20huge scale.
- 7:21>> Yeah, we also just had this blue wave
- 7:23that was the other news before the the
- 7:25Wong stuff. Um I think it was all but
- 7:28one of these data center referendum
- 7:30bills passed. In other words, people
- 7:33hate the data centers. Um a lot of voter
- 7:36sentiment on both sides of the aisle
- 7:37against AI. I'm just curious like how do
- 7:39you think the industry screwed this up
- 7:41so badly in telling the story about this
- 7:43technology?
- 7:46I I think if the sales pitch is like we
- 7:48are going to offer you a cure to cancer
- 7:51and cheap stuff and the price of that is
- 7:54you are going to lose all agency and
- 7:56control and voice in the future. You are
- 7:58going to live like a Labrador. Maybe the
- 8:01machines will try to make you happy, but
- 8:02you certainly won't have any ability to
- 8:04do that. and we're going to like rush to
- 8:05hand over um you know control of the
- 8:09future to AI and we're going to like
- 8:12make our make our AI really nice uh and
- 8:14he's going to take good care of you but
- 8:15that's going to be that. I think it's a
- 8:17terrible sales pitch. I don't want that
- 8:19world. Um I and I don't think most
- 8:22people want that world. There's other
- 8:24big questions about the economics too.
- 8:26But I I think fundamentally if you talk
- 8:29to people who really don't like AI and
- 8:31you get past the initial things. I think
- 8:34data centers are easy to point to
- 8:35because they're like a big physical
- 8:37instantiation of this broader thing. I
- 8:39think the job stuff it's it's a real
- 8:42immediate term worry that people feel
- 8:44very deeply. But there is this like
- 8:46long-term question about what are
- 8:47people's role going to be in society and
- 8:51what you know what is the balance of
- 8:53power going to look like either between
- 8:55people and AI or between people that
- 8:57have a lot of AI and a lot of computing
- 8:59power to use AI versus people who don't.
- 9:02Um and I think this is a place where the
- 9:05AI industry needs to take a very strong
- 9:06stance which is we are
- 9:09aggressively in favor of and committed
- 9:11to people staying in control of the
- 9:13future. um we should not want uh to have
- 9:17a handover, I think that's the current
- 9:19term people are using in Silicon Valley.
- 9:21Uh either gradual or sudden to AI uh you
- 9:23know making decisions about our future.
- 9:25This should be a tool that people get to
- 9:26use and
- 9:29I think people can kind of tell that not
- 9:30everybody in the AI industry thinks that
- 9:32way and are appropriately terrified. uh
- 9:34and that combined with the economic
- 9:36uncertainty and you know various ways
- 9:38that AI has you know caused cyber issues
- 9:41or whatever um has people against data
- 9:45centers but I think from the
- 9:48conversations I've had that what they're
- 9:49really against is this feeling that man
- 9:52are the people building AI actually
- 9:55really truly on team human. You said
- 9:57back in uh July 2025 you made this
- 10:00comment about Democrats uh you're
- 10:02politically homeless. Democrats have
- 10:04lost the culture of innovation. Do you
- 10:06still feel that way?
- 10:09>> Um
- 10:11roughly speaking, you know.
- 10:13>> How does that affect your view on the
- 10:15the the results, you know, what we're
- 10:18seeing now? I mean, obviously I will I
- 10:21want to work with whoever our elected
- 10:23leaders are. Do I still think there is
- 10:24room for like a party that more closely
- 10:28matches
- 10:30the ideals that I think have worked for
- 10:32America and the spirit of innovation for
- 10:34a long time. Um, I think that'd be
- 10:37great. I hope that both parties will
- 10:39move in that direction and I think there
- 10:40are like positive signs in both
- 10:43directions. Um,
- 10:46but yeah, you you said you want to work
- 10:48with the administration. you've, you
- 10:50know, before. I mean, you could call up
- 10:51David Saxs, you could call up Susie
- 10:53Wilds, can you call up Steven Miller?
- 10:56>> I don't know. I'll try.
- 10:59Have you, Do you still, you know, just
- 11:01curious like when was the last time you
- 11:03talked with Trump?
- 11:05>> Um,
- 11:08the China State Dinner.
- 11:10>> Yeah. And And what was the discussion
- 11:13then? What did you guys talk about? I
- 11:14>> mean, you don't get to have like a deep
- 11:16in-depth conversation. That's a lot of
- 11:19stuff. But yeah.
- 11:20>> So, you know, I can't help but think
- 11:22about you had your own spy drama, right,
- 11:24with this, you know, person putting a
- 11:26hidden uh a hidden microphone in your
- 11:29office. I think you've talked to
- 11:31employees about the the danger of, you
- 11:33know, people pretending to hit on them
- 11:36and potentially being agents of a
- 11:38foreign government. Um, you know, I'm
- 11:40just curious like has this changed how
- 11:42you interact with you have a lot of
- 11:44investors from the Gulf. Do you view
- 11:46them differently because of all this
- 11:48stuff?
- 11:50I
- 11:52we have now had I think more practice
- 11:55than is maybe obvious from the outside
- 11:57about living in this world where it is
- 12:00clear that AI is an extremely important
- 12:03thing happening and that people are
- 12:05looking for an edge on in all sorts of
- 12:08ways whether it's you know putting a
- 12:11camera in one of our conference rooms or
- 12:13trying to take our engineers out for a
- 12:15few drinks and get them to tell
- 12:16pre-training secrets um or whatever.
- 12:18ever. So,
- 12:21I think we have been living in this
- 12:24reality that everybody else is now kind
- 12:26of very publicly and suddenly coming to
- 12:28grips with. Um, and
- 12:32I'm sure we've been tricked sometimes
- 12:34and I'm sure that our operational
- 12:37security and overall security has not
- 12:39been perfect, but we have had time to
- 12:41adjust to this. the the a thing that we
- 12:44talk a lot about uh when it comes to
- 12:46safety in which I put this bucket of of
- 12:48stuff like the I put security in the
- 12:50broader bucket is being pragmatic
- 12:52centrists. Um it it totally does not
- 12:56work to I think say we're going to just
- 12:59build AI in absolute secret in a lab.
- 13:02We're not going to ever trust anybody
- 13:03else with it. We're not going to bring
- 13:04the world along. We're not going to
- 13:06iteratively deploy this technology and
- 13:07get the world to, you know, be able to
- 13:10use it. Um, but also we're not going to
- 13:12pretend like this is like any other
- 13:14technology. And there are no moments
- 13:15where we have to do things maybe
- 13:18differently than any kind of company or
- 13:19any industry has come before. And that
- 13:22can range from things like being very
- 13:24careful in our external interactions to
- 13:26saying, you know, at some point this
- 13:28technology may get so powerful so
- 13:30quickly that we really have to change
- 13:31all of our our plans uh to be able to
- 13:33continue to make the safety guarantees
- 13:35we want to make. And um that thing we
- 13:38talked about earlier I think this
- 13:39principle of people staying always
- 13:40standing control of the future uh that's
- 13:43going to require some very different it
- 13:45may require some very different actions
- 13:47as these models get so capable. People
- 13:49talk a lot about recursive
- 13:50self-improvement. Easy to say those
- 13:52words. I don't think most people like
- 13:53really have wrapped their head around
- 13:55what that would mean and what we might
- 13:56have to do. Um so I think for all of our
- 14:00interactions we try to like have
- 14:04the weight of what is happening and what
- 14:06we are doing on our mind in every way
- 14:09and then act as these pragmatic
- 14:11centrists where you have to kind of like
- 14:14engage with the messy reality of the
- 14:15world.
- 14:16>> Since you mentioned RSI are you there
- 14:17yet? Is that happening? Um
- 14:22I think like many other things including
- 14:24the term AGI it is like very hard to
- 14:27have a clean demarcation and say now we
- 14:29didn't have AGI now we do. Um are we
- 14:31developing
- 14:34new models faster than we would without
- 14:36using our technology? Yes. Are we
- 14:39pushing a button and just saying hey
- 14:40model like
- 14:43design the next version and rapidly
- 14:45self-improve with no human oversight?
- 14:46Definitely not. um we could be pushing
- 14:49more towards the RSI spectrum than we
- 14:51are, but I don't think we could safely
- 14:52do that right now. And again, I'm not
- 14:58I I think this like question of if we
- 15:00make all decisions
- 15:02first from a principle of saying people
- 15:04need to be in control of the future
- 15:06forever. There are going to be a lot of
- 15:08follow-on implications from that for
- 15:10things like how quickly someone should
- 15:12be willing to push onto the R.
- 15:14>> So, you're going to hold back on RSI a
- 15:15little bit because of that. I I think
- 15:18that holding back on RSI is a tactic. Uh
- 15:22I think RSI can be a good thing.
- 15:23Certainly using this research, using
- 15:25this technology to do better and faster
- 15:27alignment research is a great thing. Um
- 15:30but I think you should model us as we
- 15:31will make our decisions um from a place
- 15:34of being unwavering on people needing to
- 15:38stay in control of the future and
- 15:41sort of this pragmatic centrist view of
- 15:43safety and interaction with the world.
- 15:45Meter uh had this report I'm sure you
- 15:47saw about Anthropic saying that they
- 15:49they've doubled the speed of their
- 15:51capability development on specific tasks
- 15:53compared to like a couple years ago. Are
- 15:55you also seeing that rate of increase?
- 15:59Um
- 16:00I don't know of a measurement like quite
- 16:02like that. Uh and and I think it'd be
- 16:04like it it is hard this these things are
- 16:08still jagged enough that it's hard to
- 16:09say like overall we're doing this whole
- 16:11class of things two times as fast. I'm
- 16:13sure you could find some things we're
- 16:14doing much faster than that, some things
- 16:16that are much slower. The blended
- 16:17numbers I think are always harder than
- 16:19they seem. Um, we are speaking of Meter,
- 16:22uh, we are working with embedded
- 16:24evaluators now, uh, including meter. And
- 16:28I think that this principle of having
- 16:31people
- 16:33getting like an outside opinion on the
- 16:35safety work and the sort of robustness
- 16:37of these systems is going to be
- 16:40increasingly important and valuable.
- 16:42So, let's talk about products. I use
- 16:44dots. I find it incredibly useful. Um, I
- 16:48was looking at the terms of service and
- 16:49I was sort of wondering like where's
- 16:51where's my data? Like where's this
- 16:52memory being stored and it's somewhere
- 16:54on the cloud. Um, currently I can't
- 16:57change it or you know edit it in any
- 17:00way. Um, and I'm just curious like it
- 17:03worries a little bit worries me a little
- 17:05bit about like lock in. would you allow
- 17:07people to sort of have a mechanism to
- 17:09export their dots uh memory to other
- 17:13>> so we do for chipt memory um and that's
- 17:16like an easier structure the the
- 17:20I think in principle yes people should
- 17:23be able to take their data with them the
- 17:25dots version of this is harder than the
- 17:30chatebt version of this like dots really
- 17:33can build up like a quite sophisticated
- 17:36understanding
- 17:37of your life. Um, so I don't have like a
- 17:40ready-made answer. I'll say I I agree
- 17:42with the principle in general of like
- 17:43people people should have their data and
- 17:45be able to do what they want with it.
- 17:46But it do you think people will be sort
- 17:48of worried about that? It's like a when
- 17:50you talk about this this thing living in
- 17:52the cloud that's like got a full
- 17:53understanding of my life. It just seems
- 17:55like I want to sort of own that. I don't
- 17:58know. I don't want that to be out in the
- 18:00cloud and I have no control over it. I
- 18:02can't even see it. Well, I think you
- 18:03want some way to be able to see it, to
- 18:06modify it, and to take it if you want to
- 18:08take it. Uh, exactly what that look
- 18:09like. It's easy to think about, you
- 18:12know, and someone said to me recently, I
- 18:14thought it was a good insight that the
- 18:15biggest lock in the entire tech industry
- 18:17is your um your iPhone camera roll and
- 18:20your eye messages. you just like that is
- 18:23over now 20 years of our lives and it's
- 18:25there and you really like you don't
- 18:26think about it because you can kind of
- 18:28sit there and look at it and probably
- 18:30theoretically you could export it but
- 18:31you're not going to. Um and
- 18:35but that's at least like you can just
- 18:36say like I have 100,000 photos I'm going
- 18:38to move them somewhere else feels
- 18:40theoretically plausible. Um, if there is
- 18:42just this like
- 18:44subtle understanding in the same way
- 18:46that if you have like a really great
- 18:48assistant that understands your life and
- 18:50your preferences and that like oh he
- 18:52normally wants this but in this weird
- 18:53circumstance he wants this. Those are
- 18:54like hard rules to write down. I mean
- 18:57maybe maybe dots can do it in a way that
- 18:59person that people can't. Um and again I
- 19:02agree with you on the principle. I just
- 19:04like haven't thought before this about
- 19:05like what the actual button in the
- 19:07product version of this looks like.
- 19:08>> Well, you never want to fire that
- 19:09assistant. That's for sure. Right. So
- 19:11that's yeah
- 19:13>> good good customer lock in there. Um
- 19:15okay so speaking of consumers and
- 19:18everything you got in some hot water
- 19:20talking about computer RAM uh the other
- 19:22day. Um you know you said I think it's a
- 19:26bad allocation of resources for people
- 19:27to have all this RAM on their home
- 19:29computers. Um you know people feel like
- 19:32I should be able to tinker with this
- 19:33stuff. I should be able to use it on my
- 19:36you know run models on locally etc. um
- 19:38that was a big part of like the PC
- 19:41internet revolution was like these home
- 19:44tinkerers who do cool stuff in their
- 19:46home computers. Do you think that we're
- 19:48like we're missing that in this era of
- 19:50cloud everything in the cloud? I
- 19:51>> I want to be clear. I want people to
- 19:53have a lot of RAM. I think everybody
- 19:55should buy a lot of RAM, be able to buy
- 19:56a lot of RAM for their own computer. I
- 19:58think this is good. I think the home
- 20:00tinkering is great. Um unfortunately a
- 20:03few years ago we started saying like the
- 20:05world needs to build a lot more
- 20:08memory supply chain in particular. We
- 20:10said the whole chip supply chain, logic,
- 20:12memory, packaging, whole thing. We said
- 20:14in particular there was going to be a
- 20:15huge squeeze on memory coming and you
- 20:18know like someone quoted me as saying
- 20:21the world needs to invest $7 trillion in
- 20:23capex and for like I was in much more
- 20:27hot water for that than the RAM comment.
- 20:29Everyone was saying this is ridiculous.
- 20:31This is impossible. This is insane. The
- 20:32world will never do this. This is
- 20:34ludicrous. this is, you know, there's no
- 20:36there's no way there's this much demand
- 20:38in the world and the world did not build
- 20:40out that much and now we have seen DRAM
- 20:43prices. I don't know exactly but maybe
- 20:45they went up like 5x 7x something like
- 20:47that. Um I think it would be much better
- 20:49if the world just had a lot of memory.
- 20:54um that comment which I think got
- 20:57misinterpreted but there what there is
- 21:00like a true part of it which I will
- 21:01stand by is that if we're going to be in
- 21:03this memory supply shortage then in some
- 21:06sense whether or not we'll decide to do
- 21:08it because I think there are other
- 21:09trade-offs like people having their own
- 21:11you know at home inference the optimal
- 21:14thing would be can we share this memory
- 21:16and have it working 100% of the time
- 21:18rather than like sometimes off in
- 21:20someone's edge device uh again optimal
- 21:22use of this limited supply maybe is not
- 21:25the reason to make the decision but
- 21:27that's what I was trying to say
- 21:28>> what what's the argument I mean we have
- 21:29like you know the CEO of ASML here right
- 21:31like what's your argument when you meet
- 21:33with these people like because they're
- 21:34saying I can't do I can't increase
- 21:37production really
- 21:38>> uh no I think now people are increasing
- 21:41production like I I think we will see
- 21:43production get ramped up a lot it just
- 21:47it would have been good to do it a few
- 21:48years earlier
- 21:49>> do you think they did sort of drag their
- 21:50heels on that they didn't they didn't
- 21:52believe that the this was going to
- 21:54happen. The way that someone explained
- 21:56this to me is that being in the memory
- 21:58business has been approximately the
- 22:01following story for the last several
- 22:02decades, which is you invest $20 billion
- 22:05and you make $19 billion
- 22:08and that's it.
- 22:12And so and like you know there's booms
- 22:14and busts and some years you make money
- 22:16but on the whole um it has not been a
- 22:19great business and now all of a sudden
- 22:21it is. But uh I I like the way that I've
- 22:26explained it is the way that it's been
- 22:28explained to me is that the industry has
- 22:30just been through too many booms and
- 22:31busts and so people are nervous to or
- 22:35they were nervous to really make a very
- 22:37bold capex commitment.
- 22:39So on the data center point, do you
- 22:42think there is all this opposition? Will
- 22:44the data centers still get built or is
- 22:46this going to really hamper your ability
- 22:48to to do the buildout?
- 22:51Um I think data centers will get built.
- 22:53Exactly where exactly how you know what
- 22:56what the things that make sense for the
- 22:57communities they're in or whether we
- 22:58kind of put them further out away from
- 23:00communities,
- 23:03that stuff I I think wins and public
- 23:06sentiment will change over time. Um it
- 23:09wasn't that long ago that everybody
- 23:10wanted a data center in their community
- 23:13for the jobs and the kind of industry
- 23:14and you know like it was now clearly
- 23:17they don't people like put them way out
- 23:19in the desert. Uh maybe people want them
- 23:21in their communities again. Um I I
- 23:24personally have like a lot of sympathy
- 23:26to not wanting to
- 23:28have a data center in a community but I
- 23:30also think there's a lot we have a giant
- 23:32country. We have a lot of space we can
- 23:33do this. And I think people really want
- 23:35to use AI. People also want a lot of the
- 23:37other things that data centers provide.
- 23:39People like the internet and like
- 23:40watching videos on their phones or
- 23:42whatever. Um, so I think data centers
- 23:44will get built, but clearly the current
- 23:47way they're working being built and the
- 23:50engagement with communities is not
- 23:52working for people and you know the
- 23:54ecosystem will adapt. Do you think this
- 23:55it's a concern? I people talk about this
- 23:58buildout being sort of funded by debt
- 24:01and that it's all a house of cards. Do
- 24:03you do you think the debt is a problem?
- 24:06Um
- 24:08I have never seen the actual revenue
- 24:12growth
- 24:14including great margins show up for any
- 24:18revolution in tech that I have witnessed
- 24:20in my career. If it were not for that uh
- 24:23then I think it would be a problem. And
- 24:24I also think that you know these are
- 24:26like smart people on the whole issuing
- 24:28the debt. Um there's going to be like in
- 24:31any moon, there's going to be a lot of
- 24:33foam on the top that maybe needs to come
- 24:35off. Um there will be dumb things here
- 24:37and there, but but the fundamentals of
- 24:43not only transformative
- 24:45things happening in business today, but
- 24:48business is now coming to us and saying
- 24:49like I have seen the light. I see what's
- 24:51going to happen here. I can see what
- 24:52this does to the cap products and
- 24:55services I can offer in the future. um
- 24:59that just seems
- 25:02extremely clear and compelling. Now, is
- 25:06it possible that the buildout can get
- 25:09ahead of how quickly the economy can
- 25:11digest it? Yes, that could happen. Um I
- 25:13don't see that level of recklessness
- 25:16happening right now. Um the and I also
- 25:21think we're just we're underpricing
- 25:25we're underweaiting what can happen with
- 25:26the next generation of products. You
- 25:27know we had the chat bots then we had
- 25:30the coding agents. I think in the last
- 25:32few months we started to see the
- 25:34beginning of AI for science and what can
- 25:36happen with
- 25:38using a lot of compute to discover new
- 25:41math new physics. Um, I think we're
- 25:44going to start to see what doing a lot
- 25:46of compute can do for chip design and
- 25:49other some of the most valuable kind of
- 25:51engineering projects we have. I think
- 25:53we're going to see what a lot of compute
- 25:54can do to do a lot of the kind of wrote
- 25:57work inside an organization. So, and all
- 26:00of that has great value and I expect it
- 26:03will keep going.
- 26:03>> I mean, you announced actually just this
- 26:05morning that you're going to expand free
- 26:07compute for academic researchers. Um,
- 26:10you mentioned these math experiments.
- 26:12You've you've solved these historically
- 26:14difficult math theorems. Um, I'm curious
- 26:17what you think the impact of those, you
- 26:20know, particularly the math discoveries.
- 26:22Like you're you're invested, for
- 26:23instance, in nuclear fusion companies
- 26:25in, you know, companies that are
- 26:27partially reprogramming cells, like
- 26:29really cutting edge science. Do you see
- 26:32immediate impacts or will it be more of
- 26:34a long-term thing?
- 26:35>> Some of both. Uh generally I'm I'm I'm a
- 26:38believer that
- 26:40maybe the most important driver of
- 26:42long-term human progress comes from
- 26:44better scientific understanding. And
- 26:46sometimes that happens that very quickly
- 26:50changes uh society and the economy like
- 26:55you know we figured out how to make
- 26:56language models and very quickly the
- 26:57economy changed. Um, and sometimes it's
- 27:00like a fundamental discovery in physics
- 27:02and it's going to apply in a super
- 27:05important way to a new material or
- 27:06whatever, but that will take decades
- 27:07longer. So, I think it will be a mix.
- 27:09But what we're hearing from scientists
- 27:11is so compelling now about how much more
- 27:13productive they are. And not just that
- 27:14they can work faster, but they can
- 27:16discover things that they just weren't
- 27:18going to be able to figure out
- 27:19otherwise. We heard this a lot from
- 27:20mathematicians. We're now hearing it
- 27:22from physicists, chemists, material
- 27:24scientists. um and a world where we can
- 27:28get
- 27:2910 years or 100 years of science done in
- 27:32a year. Um and then the ability to
- 27:36compound that into valuable products and
- 27:39services and technologies we all use. Um
- 27:42I think the world can get so much better
- 27:44if we think backwards
- 27:46100 or 500 years and what it was like to
- 27:48live then. I think we are very grateful
- 27:51for the scientific discoveries and the
- 27:53engine to get them usable in people's
- 27:55hands we have had since um and I think
- 27:58in the notistant future I hope people
- 28:00will look back at 2026 and be like they
- 28:02had it really rough and the main reason
- 28:04will be the scientific discovery and the
- 28:06compute and the AI that is being used to
- 28:08and the robots whatever to make our
- 28:10lives uh just much more enjoyable higher
- 28:13quality more fulfilling higher agency to
- 28:15live. Um, so we were thrilled to give AI
- 28:19uh to the scientists and I think that
- 28:21the next couple of generations of models
- 28:23are really going to surprise people on
- 28:25the upside of how much they can be an
- 28:27engine of scientific discovery.
- 28:29>> Well, I think we're going to end it on
- 28:30that optimistic note. Okay, it's a good
- 28:32one. I really appreciate the time. Thank
- 28:33you very much.
About this transcript
This page contains the full transcript of Sam Altman: ‘It’s Hard To Imagine the AI Revolution Without Jensen’ | Silicon Valley & The World by Semafor, generated from the public captions YouTube serves with the video. The transcript has 5,324 words across 766 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.