
The Economist's Zanny Minton Interviews Dr. Eric Schmidt on Technology's Geopolitics After #AI
What this covers
This is a 15-minute onstage conversation at the Future Investment Initiative in Riyadh. Zanny Minton Beddoes of The Economist interviews Eric Schmidt, the former Google chief executive. It begins with how AI reshapes national power and then moves to warfare, regulation and Saudi Arabia's prospects. Schmidt explains why the American financing model produces AI winners. Companies there can build ten billion dollars of hardware before they know how to earn revenue from it. He also stresses that the binding constraints are electricity and data center capacity, not just chips.
Most of the substance is military. Schmidt describes Ukraine as a live laboratory where each side copies the other's tactical ideas within three to six weeks. He notes that a $5,000 drone can destroy a $5 million tank. He lays out three stages of drone warfare: a person targeting a single drone, swarms against one target, and automated readiness systems that launch drones in response to detected threats. He says Russia, historically slow to innovate, has built large drone factories and drawn on technology from Iran. He argues that militaries are organized around the wrong assets, such as aircraft carriers and tanks, and that wartime pressure overrides regulation and international agreements. He also observes that cheap drones give small states and non-state actors capabilities they never had before.
The more speculative turns concern deterrence and AGI. Schmidt says it is unknown what happens when two sides hold autonomous systems on a hair trigger, and that the West has no doctrine comparable to Mutual Assured Destruction (MAD)). His most contested scenario has China reaching AGI first in a secret lab, then finding weaknesses in how banks communicate and launching a zero day cyber attack. He argues the West must get there first to set the agenda. The Saudi-specific advice is that the kingdom should use its power, sun, capital and connectivity to host training and inference that Europe and Britain cannot afford. He claims the US will run short of electricity for AI by 2028.
AI represents a geopolitical revolution larger than the PC or internet revolutions, with the US and China as clear leaders, but the critical question is how the other 195 countries will be organized, and future warfare will be dominated by autonomous drone networks requiring fundamentally new deterrence doctrines.
- AI revolution is larger in scale than PC and internet revolutions because it concerns intelligence itself
- The organizing principle for winners will be access to capital, energy, and data center infrastructure, not just computing power
- Warfare is already evolving in Ukraine at 3-6 week innovation cycles with drone swarms replacing traditional military assets, establishing a template for future conflicts
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
The American model is uniquely capable of producing winners in AI because the US has deep financing scenarios allowing companies to build 10 billion dollars worth of hardware without necessarily knowing how to generate revenue from it
“The American model is uniquely capable of producing winners in this particular scenario because we have such deep financing scenarios that people can build literally you know 10 billion dollars worth of Hardware without necessarily any clue as to how to make any revenue for it um which is sort of a gift from God as far as I'm concerned”
There is an enormous shortage of electricity in the developed world, with Europe and Britain having electricity costs so high that large-scale AI training and inference operations are economically infeasible there
“there is a huge shortage of electricity in the developed world the US 1s is the US will run out of electricity for the kind of stuff I Do by 2028 um as you know Europe and Britain the electricity costs are so high”
A $5,000 drone can now destroy a $5 million tank, making tanks largely useless in modern warfare and creating an economic incentive to replace tank-based military forces with drone-based forces
“one has sea dominance because of sea drones and that's Ukraine the the front line is about 10 km wide normally sometimes it gets down to half of a kilometer and those things are death zones they're the most mined uh facilities in the world they're mined because both sides don't want Vehicles going back and forth um as you all know the world has an awful lot of Tanks um those tanks are largely useless now a $5,000 drone can destroy a $5 million tank um I read somewhere that the US had thousands and thousands of Tanks stored somewhere give them away right buy a drone instead in fact buy 10 buy 20 buy 50 buy 100”
Both sides in Ukraine are innovating at incredible speeds, with any tactical idea adopted by the opposing side within 3 to 6 weeks because both sides are watching each other closely and rapidly implementing successful innovations
“the way it really works is both sides are at at incredible speeds innovating and the general rule in Ukraine is that any idea on one side is adopted by the other side within three to six weeks because they're both watching each other um and Russian tactics have recently changed to copy Ukrainian ones Ukrainian ones have recently adopted some Russian ones and so forth”
These three-stage drone tactics (manual targeting, swarming, autonomous air defense) are based on US Air Force tactics developed 70 years ago in World War II, and Ukraine is already seeing all three stages with Ukraine ahead of Russia but not by much
“those are the tactics that the US Air Force developed you know 70 years ago in World War II you're seeing that now in drones today Ukraine is ahead of Russia but not by much in those tactics”
In any future conflict, both sides will end up with automatic networks of drones, creating a new deterrence problem: it is unknown what happens in a deterrence regime when both sides have autonomous systems on a hair trigger, and the West does not yet have an equivalent doctrine to the Cold War's Mutual Assured Destruction
“I think a reasonable conect conclusion is that in any future conflict both sides will end up with that outcome and I should say that I don't know what happens in a deterrence regime when both sides have automatic networks of drones um if you look at the literature and the thinking people have not thought through what it means to have those systems on a hair trigger this is the new deterrence problem and in in Henry or Dr uh memory he pioneered all of his thinking uh in the 1950s around Mutual assur destruction we don't have an equivalent Dogma equivalent Doctrine yet”
Drone tactics in Ukraine are changing every day, and commanders must be in the field because the war is changing so quickly that rear-area planning becomes obsolete
“the Drone tactics are changing every day when you talk to the commanders they say you actually have to be in the field because the war is changing so quickly”
Russia, historically non-innovative, has significantly changed its approach to warfare under the pressure of the Ukraine conflict, building huge drone factories, incorporating technology from Iran, and perfecting it, demonstrating that wartime pressure forces even traditionally bureaucratic militaries to innovate rapidly
“Russia which has historically been non-innovative would be an under understatement um have under the enormous pressure of the the war that they started they really are changing their own definition of War they built these huge factories for drones they took technology from Iran and others and and have perfected it”
The key question is not what happens in the US and China, but what happens to the other 195 countries, as it is not at all clear how they will be organized in the AI era given that large pools of capital and energy and the ability to build data centers will be key determinants
“to me the key question is not what happens in the US and China because we can talk about that it's pretty clear but what happens to the other 195 countries right it it's not at all clear to me how that organized is uh software in general is inexpensive easy to use hard to control and distributed very broadly in this particular case of AI the presence of very large pools of capital and energy and the ability to build the data centers will be a key Point”
Military organizations are structurally wrong for adapting to AI and autonomy because the entire universe of what you do is changing while the organizational structure, priorities, and political structures are not changing
“I'm somewhat cynical about military people and organizations because I spend so much time with and the reason I'm cynical is not because the people are bad they're very very good people but because the system is precisely wrong for this kind of a scenario the entire universe the entire floor of what you do is changing and yet the organization the organizational structure the priorities the political structures are not changing”
In future conflicts, technology advances will favor small, mobile, autonomous devices like drones over slow, static, long-term assets like aircraft carriers, but militaries are not organized around this reality and continue to prioritize traditional asset platforms
“it's obvious if if you're a technology computer scientist that future Wars will be used with devices that are at triable uh multiple U very very shortterm and very very mobile things like aircraft carriers are exactly the opposite of that right so when people like me think about this we think about why don't you in fact build a drone Army and deal with things autonomously”
AI regulation by governments will be very difficult because governments cannot follow what's happening in technology, and governments cannot adopt AI effectively unless they operate secret labs that function differently from current military structures
“my view on the AI regulation question is that um it will be very difficult for governments to to follow what's going on but it will also be very difficult for governments to adopt them unless they have secret labs and so forth which operate differently from the current military”
In actual warfare, regulatory frameworks and international agreements do not determine innovation behavior—combatants innovate as fast as possible to stay ahead of enemies regardless of rules, as evidenced by the pace of AI innovation in Ukraine
“when you are actually at War and this is what struck me in Ukraine it doesn't matter what the rules and regulations are you are going to innovate as fast as you can to stay one step ahead of the enemy and this is there's unbel believable pace of AI Innovation happening there”
Because marginal costs of drones are very low, small states and non-state actors now have military capabilities (autonomous drone networks) that were previously unavailable to them, fundamentally changing the military landscape at both the superpower and sub-state level
“you have that at the level of superpowers but you also have because the marginal cost of these drones is very low you have capabilities for small states or indeed non-state actors which are very very different to what they've had in the past so at both ends Warfare changes”
Both the US and China are going to be clear winners in the AI revolution, but the geopolitical outcome for both is less important than understanding how the other 195 countries will be organized
“I think it's fair to say that both countries are going to be clear winners in this Revolution”
The same transformative power of AGI that creates opportunity for tremendous positive outcomes also creates symmetrical threats—asymmetric access by terrorists or adversaries to AGI capabilities could grant them enormous power and is a critical national security issue for the future
“the same thing is also true for threats against us so for example access to AGI by a terrorist how do you make sure that a terrorist group cannot get such enormous power in an asymmetric way into their hands these are the great National Security issues for the future”
If China developed AGI first in a secret lab with Chinese values and rules very different from Western Democratic views, it could redesign the world's financial system, discover bugs in how banks talk to each other, and create zero-day cyber attacks on cyber infrastructure—outcomes the West cannot predict or defend against
“imagine that there's a secret in lab in China that invents something with that kind of flexibility well ahead of the West just imagine that and remember it would have Chinese values Chinese rules which are very different from Western Democratic view uh uh views um in that scenario what could occur so this is a system in our fiction because we don't know it yet that could for example imagine how to change the world's Financial system and the way the banks work it could figure out a bug in the way the banks talk to each other um it could for example um suggest a new they're called zero day cyber attacks these are unknown attacks on Cyber infrastructure we just don't know”
To win in AI, Saudi Arabia should use its abundant resources to create data center facilities that no one else can build, because there is a huge shortage of electricity in the developed world—the US will run out of electricity for AI training by 2028, and Europe and Britain have such high electricity costs that training and inference can be done in Saudi Arabia with the right networks and data centers
“The simplest thing for uh this country to win is to use its abundant resources and create facilities that no one else can build there is a huge shortage of electricity in the developed world the US 1s is the US will run out of electricity for the kind of stuff I Do by 2028 um as you know Europe and Britain the electricity costs are so high there's every reason to think that the training and the inference can be done here with the right networks the right data centers and so forth and that's what I think all of us want this country to do”
Saudi Arabia could be one of the great winners in AI if Saudi does the right things, because it has abundant power, clear leadership, correct financing, correct connectivity, favorable weather, and lots of sun for renewables—every reason to think Saudi can become one of the big winners if it puts the money that's in place now wisely and quickly
“Saudi by the way could be one of those great winners if Saudi does the right things but but let's just focus in terms of National Security and geopolitics”
The cost of autonomy has fallen so quickly that the future of conflict will eliminate tanks, artillery, and mortars, not because conflict goes away but because these asset classes will be replaced by autonomous systems
“the cost of the cost of autonomy has fallen so quickly that the Drone War which is the future of conflict will get rid of eventually tanks artillery and mortars I'm not suggesting by the way that a conflict goes away I just think it changes”
Future warfare will be dominated by drone networks operating in air and sea (with land drones coming), where the first stage is a man targeting a single drone, the second is a flock or swarm of drones against the same target, and the third is a readiness system where drones are autonomously ready and the computer launches them in response to detected threats
“the simplest way to understand it is that the Drone World think of them as a robot world so if I say robots you become afraid because you saw those movies drones are much more deadly right so this drone world air and sea right now land drones are coming and I've seen them in use uh are in fact in a network so the steps are first a man a Target and the use of a drone to Target it the next one is a flock right a swarm technically of drones launched against the same Target the next step after that is a Readiness system where the drones are ready the computer figures out if the threats coming launches the drones and they serve as air defense”
The West and particularly the United States needs to invent AGI very quickly because getting there first allows you to set the agenda, even though it does not mean an opponent like China is vanquished
“all of us who work on this believe very strongly that the west and in particular United States needs to invent this very very quickly it is in fact a race it the fact that we would get there first does not mean that somehow our opponent in this case China is vanquished but it does mean that you can set the agenda”
The arrival of a new intelligence that is humanity's partner, making each person as brilliant as the best physicists, chemists, and artists, will change all lives forever within our lifetimes, with enormous positive implications for health, education, and human development
“enormous optimism because the arrival of a new intelligence that is our partner that makes each and every one of us as brilliant as the best physicists the best chemists the best artists and so forth our lives will be changed forever and it'll occur in our lifetimes the optimism in terms of health uh health education the development of your children and your grandchildren is enormous”
Both enormous optimism and enormous concern about AGI are correct because the same technological breakthrough creates both unprecedented opportunity and unprecedented risk
“I think the correct thing is en to say it is you have enormous optimism and enormous concern and they're both correct”
The AI revolution is of a scale larger than the PC Revolution and larger than the internet Revolution because it is fundamentally a revolution about intelligence itself
“this revolution is of a scale larger than the PC Revolution as larger larger than the internet Revolution because it's a revolution about intelligence”
In the Middle East, easily targeted non-moving assets like oil infrastructure and critical facilities can be threatened by drones, making National Security in the region dependent on drone-based air defense systems
“my understanding is that that many of the countries in this region are in conflict with Iran and so if you think about the most valuable assets here they're easily targeted by drones right because they're not moving we know where they are Etc and indeed there were Iranian attacks in this country some years ago so a reasonable expectation is that all of the National Security components within within these regions will ultimately be protected in a military sense and I think usually the correct answer will be drones and all sorts of different kinds”
The US will run out of electricity for AI training by 2028
“the US will run out of electricity for the kind of stuff I Do by 2028”
AGI is conceptualized as intelligence that has the flexibility, creativity, and learning adaptability that humans have but current computers do not, and it is coming very quickly
“there are many different definitions of artificial general intelligence uh AGI is the abbreviation but think of it as the intelligence that we have that computers don't today have it's a flexibility it's a creativity it's a oh today I want to learn something new kind of thinking which today the computers cannot do but it's coming very very quickly”
Western governments and militaries are not paying adequate attention to the direct national security and military consequences of AI
“there's been a lot of conversation in the last couple of days about AI governments around the world thinking about it do you think enough attention is being focused to the direct National Security and indeed military consequences um well I'm somewhat cynical about military people”
The speaker has personally been in Ukraine and observed the future of warfare in the AI era
“I have just been in Ukraine um I have seen the future of warfare in this era”
The speaker personally believes AI cannot be regulated effectively
“I actually would add another one which is I don't think you can regulate it that's my own my own view of this”
62% of the audience is already concerned about AI, 10% is not concerned, and the remainder (28%) presumably have intermediate concern
“so we if you those of you the 62% of you who are already concerned I predict we'll end the next 15 minutes Eric come and have a seat there even more concerned and the 10% of you who are not concerned”
Eric Schmidt co-wrote a second book with Henry Kissinger called 'Genesis' that comes out in November on the anniversary of Kissinger's death, about AI and geopolitics
“Dr Kissinger passed away at 10 and a half a year ago and his last book my second book with him comes out in November on the anniversary of his death it's called Genesis and it's about this”
Elon Musk recently changed his assessment to say there is a 90% chance that AI brings a wonderful future and 10% chance of dystopia
“Elon Musk yesterday changed his assessment he said it was a 90% chance that AI would bring a wonderful future and a 10% chance of dystopia”