解锁 AGI：AGI 如何改变每个人的生活｜Jack Hidary、Salim Ismail 与 Dave Blundin | EP #213
这场讨论认为，AI 与数据中心扩张受能源硬件约束，但预计大约10年内，稀缺将转向充裕。 Diamandis 对比称，中国在2024年新增429 GW电力，而美国仅新增51 GW；Hidary表示，燃气轮机交付周期已达到4.5年，但SMRs、新建涡轮机工厂和太阳能材料突破可能在7到9年内改变这一曲线。「我们正处在一个非常奇怪的、受限的稀缺时期。」
如果算力可以奔向能源，沙特据称成本低至美国1/3的太阳能，以及未充分利用的碳氢资源，可能吸引数据中心，而后者的沉没资本会让其扎根当地。 Blundin的逆向判断是，能源过去通常流向人口，而「AI可以来到能源身边」。一旦巨型数据中心建成，他认为即使未来聚变能源可用，数据中心也不会仅仅因为这一点而搬迁。
充裕能源重新定价的将是水、医疗、交通、食品以及依赖石油的主权经济体，而不仅是电力。 Hidary表示，海水淡化成本可能大幅下降，让淡水变得普遍可得；Ismail称，非洲大陆50%的医院床位被坏水相关感染和疾病占用。Hidary将加拿大每桶约50美元的边际开采成本，与海湾地区的7美元进行对比，并提出一个判断：俄罗斯油气出口可能大幅失去价值。「能源会撬动一切」（Energy tips everything）。
人形机器人最先具备投资价值的可能是工厂、物流和医院，而家庭仍是存在争议的终端市场。 Diamandis表示，5家“机器人造机器人”工厂正在建设中，其中2家在美国、3家在中国，预计将在6到18个月内完工；他还称，美国约20%的医院已经在使用基础型机器人助手。Ismail则提醒现实约束：「能不能先把 Roomba 运转起来」再预测每个房间都会有机器人？
中国的机器人竞赛既是出口战略，也是应对劳动力池收缩的必然选择。 Diamandis援引数据显示，中国机器人进口在波兰增长1,700%、墨西哥275%、俄罗斯135%、越南114%，而先进市场的进口量反而下降。Hidary预计，中国机器人最终甚至会服务于日本的老龄人口，并可能复制 BYD 对欧洲汽车制造商造成的冲击。
Hidary对2030年的量子预测包含两个方向相反的催化剂：具备商业价值的模拟能力，以及一次网络安全断裂。 量子计算既可能在亚原子层面模拟药物、材料和聚变等问题，也可能破解手机、WhatsApp和区块链使用的现有公钥加密体系。他的警告非常直接：「我以为自己已经加密了。你确实加密了——只是现在我们能读出来了。」（I thought I was encrypted. You were encrypted—just now we can read it.）
政府入股可能推高战略科技估值，却也会制造一个该小组并不信任的先例。 在据报美国政府取得 Intel 10%持股、量子股票因类似猜测上涨15–20%后，Blundin称具体投资「非常棒」，但治理先例「糟糕透顶」。Hidary更倾向于建立国家量子算力储备；除此之外，他认为 Aramco 的1.8万亿美元市值有机会走向一家5万亿美元的能源与算力公司，而 Ismail则认为，在真正的充裕时代，估值问题本身已经无关紧要。
1. 当下的能源短缺先于供给跃升
Diamandis用供给瓶颈解释这一局面：中国在2024年新增429 GW电力，美国则为51 GW；与此同时，市场已有37家获得VC支持的聚变公司、企业采购核电、SMRs、第4代裂变技术，但供给依然不足。
Hidary给出的证据来自燃气轮机市场：燃气发电约占美国电力的40%、欧洲电力的35%，但 GE、Mitsubishi Heavy 和 Siemens 的燃气轮机交付周期约为4.5年。制造商还记得上轮扩产时吃过的亏，因此仍不愿增加产能。
Diamandis讲到，一家量子数据中心建设商拿不到所需的5 GW发电机，于是买下了市场上剩余的全部3 GW机组，以及液冷系统所需的100万个阀门。这是一种防御性采购逻辑：「全部售罄。我再也不赌这个风险了。」
Hidary预计，随着SMRs、其他核电产能、涡轮机工厂、太阳能和电池陆续到位，短缺将在7到9年内逆转。太阳能转换效率目前约为27%，但材料科学和量子突破可能推动其进一步上升；在他看来，能源充裕只是延后，并非存疑。
2. 充裕电力重新布局算力与地缘杠杆
Hidary从海水淡化谈起：能源充裕将大幅降低淡水成本，可能消除全球一半的传染病；食品供应也会变得更容易，连锁反应将扩散至整个社会。
Ismail补充了医疗机制：他称，非洲大陆50%的医院床位被坏水占用，相关问题包括感染、细菌、病毒、与水有关的蚊虫以及积水。他认为下游影响将达到数万亿美元，而不只是数十亿美元；更便宜的能源还会降低运输成本，改变全球经济格局。
石油行业的调整并不均衡。Hidary表示，石油占加拿大出口的40%，当地边际成本约为每桶50美元，而海湾地区仅为7美元。他更具推测性的判断是，俄罗斯入侵乌克兰的部分原因在于：几年后，其70%的油气出口可能不再值钱。他还预计，能源充裕会降低与石油相关的侵略动机，但也承认人类可能会找到其他开战理由。
Blundin认为，沙特「把牌打得简直完美」：天然气经常被浪费，石油出口成本又可能很高，因此在当地建设数据中心颇具吸引力。讨论提到，沙特太阳能成本仅为美国的1/3；而大型太阳能和风电开发商 ACWA Power 也是一家沙特公司。
关键的基础设施逆转在于，数据中心可以迁往能源，而不是把能源运送给用户。一旦建成，数据中心庞大的资本投入就会形成锁定效应：「数据中心不会搬走」，即使聚变能源随后改变全球电力经济。
3. 人形机器人先在受控环境中变现，再进入家庭
Diamandis以 Elon Musk 预计到2040年全球将有100亿台人形机器人作为乐观情景，每台价格为2万–3万美元。他还提到，1X以2万美元或每月499美元的价格提供 NEO Gamma 机器人；最终，这意味着劳动力成本可能接近每天10美元或每小时40美分，机器人运行的是「GPT-5、GPT-6或GPT-7」级别模型，或 Grok 级别的智能。
Diamandis表示，首批5家“机器人造机器人”工厂——2家美国工厂、3家中国工厂——已经开工，可能在6个月、12个月或18个月内就绪。他认为，机器人生产机器人，再加上能够看见并采取行动、无需逐项编程的LLM，正指向「iPhone级别的生产规模」，但他并不认同 Musk 给出的具体数量预测。
近期应用并不那么光鲜：仓库、物流，以及医院内部运送床单、食品、毛巾和药品。Diamandis称，美国约20%的医院已经有某种机器人助手，并预测2年内同一场大会上会出现大量机器人，甚至可能登台展示。
Ismail认可工业领域对「枯燥、肮脏、危险的工作」存在需求，但怀疑供应链能否很快支撑10亿台机器人。家庭里有儿童、宠物，变量远多于工厂；Diamandis称，1X可能正是希望利用这种多样性来收集训练数据，而 Ismail 则用一句黑色幽默保留了风险：「它们一路上可能会伤到一些孩子。」
4. 中国推动机器人，既是产业雄心，也是生存所需
Diamandis估计，美国约有30家人形机器人公司，中国至少有同等数量；他正在追踪的约100款人形机器人中，许多来自中国。机器人进口增长最快的是技术水平较低的买方市场：波兰增长1,700%、墨西哥275%、俄罗斯135%、越南114%；韩国、德国和美国则不在这一增长序列中。
Hidary解释了背后的人口与产业压力：中国过去几十年的10%以上增长，很大程度上依赖14亿人的制造业劳动力池，而这一劳动力池如今正在收缩。中国必须自动化才能保住优势；美国则需要机器人推动制造业回流，因为美国并没有这样的劳动力池。Blundin认同这一判断。
Hidary以日本为例，提醒市场不要过早到达终点。日本凭借 Honda 的 ASIMO 抢得先机，并预计机器人将帮助照料老龄人口，而其人口将从1.45亿降至9,000万。他的预测是：「最终会是中国机器人，而不是日本机器人」（they’re going to be Chinese robots, not Japanese robots）。他预计美国会出现一定程度的共存：价格更高的 Tesla、Figure AI 及其他机器人，可能服务于供应链敏感环节和国防应用。
5. 量子的商业冲击与安全冲击在2030年前后汇合
Peter表示，Google 的 Willow 芯片达到了 Google 所称的 quantum advantage——一个营销术语——并宣布其性能较经典计算提升13,000倍。Hidary预计，到2030年将出现「可用的量子计算机」：不是终极机器，但足以产生实质性的商业和社会影响。它可以模拟癌症和阿尔茨海默病药物，或在亚原子层面设计新的碳复合材料；他预计最终会有多个赢家，因为「这不是赢家通吃的游戏」。
按 Hidary 的说法，量子传感如今已经存在：它读取心脏磁场的能力超过心电图或 MRI。由于人类无法原生理解量子输出，AI不可或缺；Sandbox AQ 的名称正是这一判断的编码——「A代表 AI，Q代表 quantum」——两者是模拟并改变物理世界的「双引擎」。
Blundin认为，量子最大的上行空间在于「非直觉、非人类的领域」：为量子机器编程，或控制聚变过程中复杂的磁场。这是一种叠加式而非替代式的AI——「它不是在取代人类」（It’s not replacing humans）——因为它让此前无法完成的工作成为可能。
Hidary预计会出现2次类似 ChatGPT 的时刻。第一次，一旦可扩展、可通过云端访问的量子系统击败现有加密技术，黑客就可能暴露政府和企业机密；那些未能保护自己的人会发现：「我以为自己已经加密了。你确实加密了——只是现在我们能读出来了。」（I thought I was encrypted. You were encrypted—just now we can read it.）第二次，量子模拟可能对聚变等离子体，以及美国国家点火装置中的激光驱动反应进行足够精确的建模，从而将聚变商业化进程提前10–20年。这也闭合了本期讨论的回路：量子计算帮助创造能源充裕，而能源充裕反过来又为更多算力提供动力。
所有权之争仍未解决。Peter提到，美国政府持有 Intel 10%股份的消息，以及量子计算股票因政府可能入股而上涨15–20%；Blundin称这些具体投资非常出色，但先例糟糕透顶，并警告国防影响可能推动政府将行业国有化、限制创新。Hidary认同政府试图挑选赢家的历史「可怕至极」，但提出建立战略量子算力储备，而非由政府持有企业。他认为，一家类似 Aramco 或 Exxon、从能源和化工扩张至算力的公司，可能成长为5万亿美元级别；Ismail则对此表示异议：在真正的充裕时代，「整个经济都会失去意义」。
What does Sandbox AQ do, and what do you think about quantum computing?
Sandbox AQ—A for AI, Q for quantum. These are ultimately the twin engines of the transformation of our society. What technology has done for us is model our world and then be able to change our world using the data from that modeling.
AI and quantum coming together gives us a new superpower, and it’s upon us in the next 4 or 5 years. 2030 is my prediction for when I think we’ll have the kind of impact that is business impact and societal impact. It’s not going to be winner-takes-all.
These nonintuitive, nonhuman domains—the AI with the training data is just as good or better there. It’s not replacing humans; it’s adding things that we otherwise couldn’t do. Once the capability is there, we’ll have an explosion of imagination.
The first is energy. The second is robotics, and we’re going to wrap up with quantum. Let’s begin with energy. The question ultimately is: What do we do with abundant energy?
Today, China added 429 gigawatts of new power in 2024, as opposed to the United States at 51 gigawatts of power. There’s an energy gap that’s significant. We’re seeing the emergence of small modular reactors, or SMRs. We’re seeing corporations going out and buying their own nuclear power sources, investing in Gen 4 fission and investing in fusion.
By my last count, there are 37 fusion VC-backed companies that are in production, but we’re still woefully and inadequately short in the United States and most of the world for energy. What are your thoughts, Jack?
Sure. Well, Peter, it’s fascinating in energy right now. We’re in a very odd, limited period of scarcity, and so we’re not seeing the big picture, which is that abundance is coming very soon. Somebody wrote a book about abundance.
Right now, if you look at, for example, gas turbines, 40% of electricity in the United States and about 35% in Europe is supplied by gas. An increasing amount here in the Gulf is moving to gas as well. What is the wait time right now to get a gas turbine—something that looks like a propeller that allows you to make electricity? I’ll take bids starting at 6 months. What’s the wait time for getting a gas turbine?
3 years. 2 years. 5 years.
It’s 4.5 years right now. GE, Mitsubishi Heavy Industries, and Siemens are the producers. There’s a 4-year wait. They’re not willing to actually make more capacity because the last time they did, they got burned, and so that’s not happening.
We have this little, odd period where there are lots of data centers, lots of AI demand, OpenAI, Stargate, all kinds of things happening, but it’s very hard to make it happen. Starting in 7, 8, or 9 years, we’re going to flip the script to abundance.
The SMRs that you just mentioned, Peter, on the nuclear side are kicking in. Other nuclear is kicking in. Various governments are about to build factories to make more turbines, and so that’s really going to be fixed as well. In terms of solar and batteries, we’re going to see a step-function change.
Right now, we’re at about 27% efficiency in terms of photons—light going to electricity—in solar panels. That’s going to move up because of material science breakthroughs and because of quantum.
I’ll give you another story that supports exactly what you said. I took a tour of the biggest quantum data center in America, and Jeff Markley [?] had just built this thing. He tried to buy 5-gigawatt generators, but they were sold out. There were none.
So he bought every 3-gigawatt generator that was left. Then he bought 1 million valves. I said, “What are you going to do with 1 million valves?” He said, “Well, look, it’s all liquid-cooled. I’m going to need the valves, but the last time I tried to build something, everything was sold out. I’m not taking that chance anymore. I’m going to buy 1 million valves and be ready.”
So any valve manufacturing company just became a unicorn overnight.
What do we do with abundant energy? Let’s go to Jack Hidary’s thesis. 7, 8, or 10 years from now, we’ve got more energy capacity than we’ve ever been able to use. What are we doing with it?
Well, I think it’s very hard to project, right? We’ve had 10,000 years of civilization based on a scarcity of energy, and suddenly we flip to abundance. We’re going to have this amazing challenge of thinking through what do we do with it?
If you have abundant energy, desalination becomes much less expensive. If you have fresh water, you take out half of all the infectious diseases in the world. The ripple effects start to propagate.
Energy tips everything.
Energy tips everything. Food becomes easy, and the ripple effects are so profound that we have to rethink all of society.
Think about the idea that, over the last 100 years, almost all the wars have been fought over oil, right? Hopefully, we should have fewer wars. We’re human beings, so we’ll find other things to fight about, but it will decrease the need for overt aggression in the way we have it now.
How many countries in the world are dependent on the petrodollar? Russia, Venezuela, obviously here in the Middle East—
Probably about 40.
And so what happens to their economies when we have abundant energy from everything else?
Huge impact. Take Canada, right? I’m Canadian. 40% of exports are oil. When we have abundant solar or other technologies, the marginal cost of extraction there is about $50 a barrel. Here, it’s about $7 a barrel. There’s no way the expensive countries are going to be able to compete on oil in the future.
The geopolitics is crazy. One thesis I have is that the reason Russia invaded Ukraine is that, in a few years, 70% of its exports of oil and gas are not going to be worth much.
Peter, just to add to Jack’s point, you mentioned it in passing, but I want to double-click on the healthcare implications. 50% of all the hospital beds occupied on the continent of Africa are due to bad water—due to infections, bacteria, and viruses carried by water, as well as water-related mosquitoes and standing water.
Once you have fresh water because of abundant energy, you suddenly free up huge amounts of the healthcare system, and people become much healthier. The implications for water, and then the downstream effect on healthcare, are already multitrillion-dollar in nature, not multibillion-dollar, in terms of their effect.
When you think about what you can do in terms of transport, transport is expensive because of energy. If you can now transport stuff from place to place, you again change global economics.
I don’t think we’re ready. I don’t think global political leaders have really taken into account what’s going to happen starting 6 or 7 years from now, when we flip from scarcity to abundance.
Closing thought, Dave, on energy.
Yeah. Well, we’re right here in Saudi Arabia. Saudi is playing the cards absolutely perfectly because the way this will play out is that the data centers are moving to where the energy is.
This is a brand-new thing in the world. Energy used to need to move to the population, hence all the tankers. The oil had to go there. But it doesn’t have to work that way with AI. AI can come to the energy.
There are huge amounts of natural gas that get wasted. There are huge amounts of oil that are expensive to export. Why not a data center here?
What’s the quote on solar energy here?
Was it you saying—no, it was someone who was telling us about it earlier. What’s the quote on solar energy, so I can repeat it? 3. So the solar energy here in Saudi is 3 times cheaper than it is in the United States.
I was going to say, ACWA Power, the largest deployer of solar and wind, is a Saudi company.
People think of Saudi as only oil. That’s not the case. Second, on energy: once the data centers are built—
You’re locked and loaded. The data center is not going to move. These are massive, capital-intensive projects. Once that’s on the ground, no matter what happens with fusion, you’re good to go.
Fascinating. And then you have Bill Gross building the energy storage system.
Bill Gross.
And then off you go there. We'll hear from him in a little bit.
Let's move to our second subject, which is the world of humanoid robotics—embodied intelligence. Last year, I was on stage here, on the main stage, with Elon talking about robots. His prediction is 10 billion humanoid robots by 2040, at a price point of $20,000 to $30,000.
We heard yesterday—and today I interviewed Dr. Robert Playter, the CEO of Boston Dynamics. Tomorrow I'm interviewing the team from 1X. Today, 1X announced their pricing on their NEO Gamma robot. They're selling it at $20,000, and they're going to be leasing it at $499 a month.
Now, this is before we've gone into production. When we get to millions of robots, tens and hundreds of millions of robots, we're going to see the price come down at these price points. Think about $10 a day, 40 cents an hour for labor, with labor running GPT-5, GPT-6, or GPT-7—or, you know, Grok superintelligence.
What do you guys think about robots coming?
Jack, just to kick off, the first 5 factories where robots are building robots have begun construction. There are 2 in the US and 3 in China. They're not ready yet; they're not in production, but these factories will be ready in the next 6, 12, or 18 months. That's where you begin to get some of the scale that Elon is talking about. I'm not sure about the number he had, but the point is directionally correct. They're at iPhone production scales.
That's right. And when robots can make robots, now we're in a whole different ballgame. The second thing you add, of course, is that the LLMs—the software side—are getting better, and that software gets implanted in the embodied robot.
Therefore, the kind of programming we thought we had to do back when a lot of us were thinking about robotics on this stage here, a number of years ago—we thought, “Wow, the coding, the programming.” No, that's out the window, because we now have an intelligence that can see something and do something.
This is a very different future where, at FII, we can make a prediction right now: at FII 2 years from now, there'll be quite a few robots walking amongst us, maybe even some on stage presenting. I think that's going to have a profound impact, initially, on warehouse logistics—a profound impact on that whole sector—and then also in hospitals.
We're seeing already a penetration of about 20%: hospitals in America have some kind of robotic helper, not for medicine and not for surgery, but for the mundane things in hospitals—bringing up the towels, the linens, the food, the medicine, and things like that.
But eventually, we're heading towards surgery, right?
We're getting there, but right now there's already adoption today. If you go to hospitals in America, you'll see 20% adoption right now. Imagine when they're better and humanoid.
I don't buy it. I have 3 points. First, I totally agree with the industrial use of robots—massive opportunities, especially for the dull, dirty, dangerous jobs.
I think supply chain issues are much deeper than people think, so I don't see how we could end up with a billion robots down the line, at least in the foreseeable future. I think it's going to take a lot longer to get there.
I am coming around to the software adaptation side. These things are going to learn very, very quickly. But as I said on our last podcast, can we please get the Roomba working, at least, before we start thinking about all sorts of other household tasks?
I think industrial uses are where we start, and we have a profound penetration there before we bring it into the home. And your idea of a robot per room—I don't think that's going to happen.
To your point, my household is the last place.
Thank you.
I agree 100%. The amount of variables in a house—dogs walking and running around, kids—there are way more variables in a house than in any factory on this planet.
Which is, by the way, why, for example, 1X wants to go in the house. They want to have enough variety where they can pick up and get data and learn, right? We've sucked up every single data point.
They may injure some kids on the way, but they're going to learn. Yes.
One thought from our last podcast: we had that little video of the robot kickboxing.
Yes. From a marketing perspective, it feels to me like kickboxing is not the first task you should show a robot doing. It should be a massage. Have it vacuuming the floors. Have it cleaning the dishes. I don't want a kickboxing robot standing next to me. That's going to be unnerving.
You want to convey a sense of safety and comfort, not UFC in the house.
Oh, my God. Dave, thoughts?
We talked about the only reason they're humanoid at all, and don't have 6 arms like you want, is because it inspires people. When they kickbox, it inspires me. This is where China is running away with the robot Olympics.
Yeah, it's inspirational. Watching soccer games played by robots is inspirational. Let's—
Talk about the China of it all.
Peter, you had some export stats.
Yeah, I do. Here's what's going on. There are 2 hotbeds of robot manufacturing. The United States has probably about 30 humanoid robot companies. We hear about the top 5 or 6, but there are many; there's a long tail here.
China has at least that. I'm putting out my humanoid robot report in about a week's time, and we're tracking on the order of 100 humanoid robots, a lot of them out of China.
It's interesting, right? As China goes out to the world—and I understand why—and provides free, open-source AI models, we're having those models adopted by country after country. In the Chinese robot space, what we're seeing is that it's the up-and-coming countries that are not technologically forward, like Poland, Mexico, Russia, and Vietnam, that are buying Chinese robots.
Yeah.
Poland is seeing a 1,700% increase in robots coming in from China. Mexico is at 275%, Russia at 135%, and Vietnam at 114%. The countries which are technologically forward—South Korea, Germany, and the US—are not importing the Chinese robots.
Yeah, I think it's important to note, though, that the Chinese economy grew at 10% plus for, what, 2 decades, largely by bringing manufacturing into China and using this 1.4 billion-person labor pool to win manufacturing wars all over the world.
And that pool is going away.
That is about to implode so badly if they don't race to robotics.
Yeah, they have to because of the population.
That makes sense to me. It totally makes sense.
Yeah. Well, that's why they got ahead. Now the US needs to bring manufacturing back. That's going to take some time, and then robotics goes hand in hand with that, because the US just doesn't have that labor pool.
So it's a very different dynamic in China versus the US, and that's why it got into the state that it's in right now.
Professor Hidary?
I think we're going to see really different variability in terms of the penetration of Chinese robots in different countries. Let's take Japan as an example. Japan had a head start in robotics. Everyone remembers the ASIMO robot from Honda.
I do. Amazing.
Okay, program closed.
It greeted President Obama.
It's an example of being too early. It's bad to be too late, but being too early is very difficult, too.
Japan's birth rate is well below the replacement rate. They're shrinking from 145 million people to 90 million people. They've always thought that robots would help take care of the older people, because there are no young people to take care of the older people. But they're going to be Chinese robots, not Japanese robots. That's the kicker.
When it comes to the States, I think there will be some adoption of robots from China, but there are going to be sensitive areas, such as supply chain and defense, where some more expensive robots—be it Tesla, be it Figure AI, and others—will also have a play. So I think they're going to coexist here.
In Europe, we're seeing BYD, on the car side, penetrate and demolish the European car companies. My prediction is the same in robotics.
Amazing.
All right, I want to take our last 10 minutes of this Moonshots podcast to talk about the quantum of it all. We just saw Google's chief scientist, Michel Devoret, win the Nobel Prize. Google is just racking up Nobel Prizes. Congratulations to our family at Google.
We just saw Google's Willow chip reach what they call quantum advantage, which is a marketing term. We can talk about what that means, but it announced a 13,000× increase over classical computing.
Let's talk about what it means, Jack, to have a quantum advantage. Talk about what Sandbox AQ does. Eric Schmidt is your chairman; you're the CEO. You have skyrocketed. You came out of Google X at a half-a-billion-dollar valuation, and you've been on an exponential growth curve since then.
What does Sandbox AQ do, and what do you think about quantum computing?
Well, thanks, Peter. Quantum is the kind of thing that both scares, shocks, and exhilarates people at the same time. We should probably make a pill called the quantum pill so that people get excited—just take the pill.
Quantum is upon us. We've gone from 5 years ago, when people were saying, “Will anyone ever really build a real quantum computer? Will it ever have any impact?” I think it's clear now that we're headed to a yes. Now the question is when.
My prediction to you on stage here is that 2030 will be a critical year, when we'll have a usable quantum computer.
What does that mean? What is usable?
Usable means there are two aspects to quantum: the positive and the negative.
On the positive side, we're going to be able to render this world model down to the subatomic level. Why is that important? You want to make a drug for cancer, for brain cancer, or Alzheimer's. You want to make a new material out of hydrocarbon feedstock and make a carbon composite to make cars lighter and more fuel-efficient.
You want a quantum computer because it is the best modeler, the best simulator, of our physical world. That's going to be great. In 2030, we're going to do that, I think, with some really good power with a quantum computer. It won't be the ultimate quantum computer, but it'll be pretty good.
On the negative side, we're going to cross a threshold around that same year, 2030, when a quantum computer will break all cybersecurity—the public-key infrastructure in your phone. When WhatsApp says “end-to-end encrypted,” that is using a protocol that quantum breaks.
If you have blockchain—anyone here have cryptocurrency? Any Bitcoiners in the room?—you want to worry about this, because eventually we're going to have to switch the blockchain to a quantum-safe program.
So 2030 is my prediction for when we'll have the kind of business and societal impact we're talking about. We're still going to have to wait until the super-scale quantum computers to do even bigger things, but 4 or 5 years from now, I think we're going to make some really good headway.
I think several companies will be involved. It's not going to be winner-takes-all. This is not a winner-takes-all game.
There's no network effect. The U.S. government is going to take all—
They're definitely a player in the game.
That's on quantum computing. It's worth mentioning, though, Peter, that quantum computing is not the only game. Quantum sensing is here today. Sensors that can see how your heart's doing because of the magnetic field in your heart, way better than any ECG or an MRI, are actually here today.
Sensing is something that maybe we can talk about on another podcast, but look it up online: quantum sensing. Fundamentally, where is this all going? We need AI to actually interpret a lot of the output of these quantum devices.
There is no human on Earth who could see a quantum sensor and read it natively. Our minds would literally blow up. AI has been fundamental in turning these from laboratory ideas into commercial products.
Sandbox AQ—we actually named our company A for AI, Q for quantum many years ago. We saw this merger coming, because this is ultimately the twin engines of the transformation of our society.
What technology has done for us—and all of us on stage are part of this—is model our world and then be able to change our world using the data from that modeling. That's what we've done in the past 70 years, and that's where we've pushed technology forward to such a great degree.
AI and quantum coming together gives us a new superpower, and it's upon us in the next 4 or 5 years.
Salim, thoughts?
I think it's huge. One area that's going to be really interesting is that almost all the computational applications are for classical computers, and we just don't have the imagination to know what we could do with a quantum computer.
People dismiss it, but I think once the capability is there, we'll have an explosion of imagination as to the types of problems we can solve. You're probably dealing with them more than anybody else right now.
In a few years, we'll all be wearing hats like yours, trying to imagine what the quantum applications of the future will be.
He's wearing that hat so his brain doesn't explode.
That's right—holding it contained in one place. I think that's where there'll be a massive opportunity: on the application side of it.
Yeah, Dave.
I think one of the coolest things about what's happening in AI right now is that when most people think about what you do with AI, they think in terms of, “Okay, here's a person doing call-center work. Here's a person doing white-collar work. Here's a person. Here's a robot walking around cleaning my house. Let's move that over to AI.”
But actually, AI is even better in these domains. Can you program this really complicated quantum computer? Can you contain a fusion reaction with a really complicated magnetic field?
These nonintuitive, nonhuman domains—the AI with the training data is just as good or better there. That's so greenfield. It's additive to humanity. It's not replacing humans; it's adding things that we otherwise couldn't do. Quantum is just one of those things.
Jack, what's it going to look like when we hit our ChatGPT moment in quantum? Is there going to be that moment where people go, “Oh, my God”?
We're going to have 2 ChatGPT moments in quantum.
The first is going to be when secrets of our society and our companies will literally be exposed by hackers on the internet. Quantum computers are not just going to be in the hands of great governments and companies. Because they're cloud-based and because those clouds are everywhere, including in many countries, hackers will have access to quantum techniques.
Once it's known how to build a quantum computer at scale, it will be rebuilt and replicated all around the world. One of the ChatGPT moments is going to be when suddenly really confidential stuff from the CIA and from governments is going to pop up, Snowden-style, on the internet.
Unless people protect themselves, that is going to happen. There'll be a wake-up call, again roughly in that year, 2029 or 2030, when people who have not protected themselves are suddenly going to be like, “How did this happen? I thought I was encrypted.”
You were encrypted. Just now, we can read it. So that's going to be one ChatGPT moment.
The other ChatGPT moment, I think, is going to be when—and this strips right back to the beginning of our conversation on energy—the number one thing that's holding back fusion power, which would give us abundance, is addressed.
One of the many things that would give us abundance is that it's very hard to model the plasma that's in a fusion reactor. It's really hard to model when, at the National Ignition Facility, you have millions of lasers going into a little dot.
Quantum computing is the first computing platform that gives us the power to model what's happening in a fusion reactor. That will accelerate us probably by 10 to 20 years. So we're fully back now, circularly, into energy abundance.
Interestingly enough, I had breakfast this morning with Lip-Bu Tan, who's the CEO of Intel. The U.S. government takes a 10% position in Intel, and the stock has been flying ever since. Those of us holding Intel options are celebrating. Yay.
We just saw a whole number of quantum-computing stocks literally go up 15% or 20% because the U.S. government is considering taking a position in them. How do you think about the government taking positions in these technology companies? Dave, do you want to kick us off?
I think they're brilliant investments, mostly with David Sacks driving the agenda, and they're incredibly valuable for the U.S. right now. I also think they're a terrible precedent, because if you think about what prior administrations might have invested in, it ranges from the most insane to beyond the most insane.
As a precedent for the way government works, it's terrible.
Yeah. But these particular investments are fantastic.
Well, also, just the defense implications mean that they'll try and do this as a starting point in nationalizing the whole thing, which would be bad.
I mean, if we have massive quantum-computational capability in the hands of private companies, will that ultimately need to be a national asset?
I think there could be a national reserve, just like we have a strategic oil reserve. You can have a national compute reserve of quantum without having to own these companies.
I agree with Salim and Dave. The history of picking winners by governments is a horrific history.
It's bad because, actually, they often turn out to be bad government investments, but also because it restricts innovation and the unleashing of entrepreneurial spirit.
Well, it's completely inconsistent, too. You have no idea what's going to happen. Just a couple of years from now, a democracy could completely invert course. Completely invert course. No one can predict the future. It's terrible to have that inside the government.
Not that democracy is working that well right now, so that's a whole other point for another podcast.
This morning, we just saw NVIDIA hit a $5 trillion valuation.
Welcome to the first $5 trillion company.
Crazy. I'll close this out with your predictions on what other industries are going to hit a $5 trillion mark here. Jack?
The energy sector. Right now, Saudi Aramco is sitting at $1.8 trillion of market cap. If Aramco and other oil and energy companies take hold of compute the way they could in the next 3 to 5 years, it's going to take a little longer than NVIDIA, but you could see a $5 trillion energy company if they take hold.
When you think energy, think of all the things around energy. In the case of Aramco, they have their unit called SABIC, which is a chemicals producer. In the case of ExxonMobil, they are an integrated chemicals producer as well.
Exxon could be that company. Aramco could be that company. It could be 3 or 5 of them. It really depends on whether the energy companies go beyond thinking about themselves just as energy and go into compute, to think about the future of the products coming out of those energy products.
That will create $5 trillion in energy.
Dr. Alex Wissner-Gross is constantly saying on the podcast, on Moonshots, we're going to tile the Earth with compute, and that's inevitable. As fast as we can create chips and as fast as we can find energy, we're going to be building compute. And that'll be true for the whole future of humanity, starting with this—
Until we break down all the planets and—
And create a Dyson swarm around our planet.
Exactly. Dyson sphere.
But anything that's directly in that value chain—
—is going to be a multitrillion-dollar opportunity. Salim,
I call the question irrelevant because once you have abundant energy, the whole economy becomes meaningless. We're debating $5 trillion. Yeah.
$5 trillion.
I call BS on the whole question.