Warning: This is a rush transcript generated automatically and may contain errors.
Asbed Bedrossian: Hello, everyone. For years, we’ve heard that a multipolar world is coming. Well, it may be time to stop asking whether it’s coming and start asking whether it’s already here. We have Dr.
Warwick Powell back in this Conversations on Groong episode to discuss this very issue. Are BRICS and the SCO, the Shanghai Cooperation Organisation, starting to build real alternatives to the Western-led systems? Is China becoming, as our guest puts it, a great enabling power? And where do countries like India, Turkey, and Armenia fit in all of this?
Then we’re going to jump into AI where things are getting even more interesting some of the people building the most powerful AI systems are now warning that development may be moving just too fast. At the same time Washington and Beijing increasingly see AI as a strategic contest that neither side wants to lose. Our guest today, Professor Warwick Powell is with the University of Queensland in Australia and the Taihe Institute in Beijing. His research sits at the intersection of digital innovation, emerging technologies, and international political economy.
Hovik Manucharyan: Thank you for staying with us, folks. We are the only independent English-language podcast focusing on the South Caucasus . We hope that you appreciate this. And if you do, then a like, a comment, a share, whatever you can spare.
And if that can be done in the form of dollars as well, or coffee, we accept that. podcasts.groong.org/donate is where you would go to do that. And it’s much appreciated. It helps keep the light on. Anyway, on with the show.
Asbed Bedrossian: Professor Warwick Powell, welcome back to the Groong podcast.
Warwick Powell: It’s great to be with you again.
Asbed Bedrossian: Yes, indeed.
Hovik Manucharyan: Welcome, Professor Powell.
Hovik Manucharyan: So, the recent SCO summit in Bishkek and the subsequent BRICS summit in New Delhi offered a chance to test whether the emerging multipolar order is producing concrete institutions and outcomes. For instance, the BRICS New Delhi Declaration reflected agreement among a diverse group of states on local currency settlement, financial cooperation, reform of multilateral institutions, and other major international issues. The SCO Declaration similarly outlined the organization’s priorities around security, trade, infrastructure, energy, and international governance.
Now, also, there have been a number of commentaries on these declarations, but we actually want to know whether any of these are producing concrete results. what is the most important tangible, and I want to underscore that word, tangible outcome from these recent summits?
Warwick Powell: I think the most tangible outcome is a continued and demonstrable capacity for a diverse number of states to find common ground in things that they clearly share concerns about, whether they be their own economic development, the interactions between countries around economic questions, whether it’s trade or investment, technology, and how that process is settled through financial means and also in terms of their ability to work with each other on questions of security. So I think that’s the concrete thing.
The important thing about things like the SCO and BRICS to understand is that you can’t judge them by using the benchmarks of traditional Cold War blocs. and I think what often happens is that many people look to these organizations as counter-blocs and whenever the nation states member states of these organizations aren’t in lockstep with each other in the same ways that you get in things like NATO for instance there is this sense that the counter-bloc is falling apart or it’s unraveling and it’s not presenting a cohesive counter-bloc. The whole point of these organisations is actually to move beyond this kind of bloc politicking.
And they can do that and demonstrably do that by focusing on the things that ultimately matter to these countries and recognising that coordinated and collaborative actions can assist them all to advance their own national interests, as well as the broader interests of each of those countries, because none of these countries takes the view that their own ability to pursue their ambitions, security and economic development, et cetera, et cetera, can take place in isolation from what happens with other countries.
And so instead of forming blocs aimed at others, what they tend to do is find ways to share information, build cross-border infrastructure, both hard and soft. And perhaps one of the things that we can talk about as we go in this conversation is the emerging focus on what I would call the soft or the invisible infrastructure, the rules, the regulations, the standards, the operating procedures, because it’s those things that ultimately create longevity in institution building.
Hovik Manucharyan: Right. How do you address this problem? Because, I mean, I see that, if I can maybe rephrase what you’re saying, is that BRICS and SCO are playing a long game geared towards more cooperation and trade. But what measuring stick can we use to measure progress?
And can you give us some examples?
Warwick Powell: Well, trade’s an obvious one. If you look at trade relationships amongst BRICS member states, they have grown substantially over the last two decades. If you look at trade relationships amongst SCO member states, similar story. Trade has been growing strongly.
If we want to look at infrastructure development, again, we have concrete evidence of infrastructure development that will facilitate both economic development within member states, but also cross-border collaboration, whether that’s in terms of maritime transport infrastructure or terrestrial transport infrastructure. All of these things are actually moving forward. If we look at the SCO, I think we also see agreements around energy development.
So we’re moving from transportation infrastructure, and there’s still work to be done there, into energy security, where member states are able to better align their energy needs and energy capabilities with the capabilities of others. So we’re getting progress, significant progress, actually, on those fronts. There are national currency-based settlement procedures and systems in place, which is working amongst these organisations already. I think often people are looking for the grand announcement as well, you know, that there’s this new big bang, you know,
Hovik Manucharyan: Yes.
Warwick Powell: the end of the dollar, all those sorts of things. That’s not actually how it works, and it’s certainly not how the member states of BRICS is approaching their work. They’re not looking for the big announcements. What they’re looking for is the practical consequences of the coordination, collaboration, and the building that they are undertaking together.
And the results of that are already in place. The fact that member states are able to settle their trade in national currencies today is evident in the fact that the infrastructure and the procedures is already in place. for a grand announcement, it is already happening. And those procedures and institutions and mechanisms will only consolidate and expand going forward.
We have institutions like the New Development Bank and also what is likely to be in the not-too-distant future an SCO investment bank playing central roles in facilitating the settlement of accounts for both investment and in trade between the member states in national currencies. Technology-wise, we have the means by which people from and organisations, obviously, from the different countries are able to make purchases in very simple and straightforward ways, whether it’s through QR code payment integrations or other platforms.
So, again, I think it’s important to recognise that a lot of the work that gets done is just work that’s done without the big announcements. There’s no big brand name or anything like that. and it wasn’t released under a BRICS umbrella or an SCO umbrella, and it wasn’t released under a BRICS umbrella or an SCO umbrella, but nonetheless is infrastructure that contributes to these things, is the CBETS platform. So CBETS platform that was released by the People’s Bank of China in around June this year.
And what CBETS does is that it creates a single interface point for counterparty banks, whether they’re central banks or commercial banks, to build digital connectivity into either the RMB payments path for the digital RMB or into the mBridge decentralized ledger-based multi- CBDC, central bank digital currency system.
And the reason why CBETS matters is that instead of these two things operating as discrete systems that require a separate set of implementations, they’re now bringing them together into a single point of interface where the technology teams from counter-party banks can move through a stage process to increase their ability to access the platforms and the services that are available from very simple cross-border trade settlement requirements, which would take about three months’ worth of technology work and governance work to settle, you know, to put into place and operate, through to accessing the full suite of cross-border finance solutions, credit, et cetera, et cetera.
And that would take an estimate of between 12 and 18 months or so of technology development, testing, implementation, So that’s actually a really important step forward, but you don’t see that as made as part of any big grand announcements.
Hovik Manucharyan: Yeah.
Warwick Powell: They’re just sort of practical things.
Asbed Bedrossian: Professor, one quick question.
Warwick Powell: So they’re the examples.
Asbed Bedrossian: So does CBETS actually enable cross-border trade in the local currencies without
Warwick Powell: Yep.
Asbed Bedrossian: making the dollarization necessary?
Warwick Powell: Yeah, so what it does is that it works on the basis of either transacting in RMB, so it gives counterparties access to the RMB payments network, or for those countries that have central bank digital currencies, and they are a new thing, so this is really about laying out the foundations for the future, that is also possible.
Let’s not forget that mBridge as a project emerged about five years ago and has now developed to a point where it is actually facilitating real transactions between companies from the participating nations. and again, as an infrastructure, it is now able to support more central banks and commercial banks integrating into it once those central banks and national governments have come to a view on how they want to operate a digital currency environment. So on both of those fronts, it provides access. Now, CBETS is obviously a very China-oriented infrastructure.
But because of its integration or its pathway into the mBridge process, it is possible now for something like mBridge or something very similar to mBridge to become the operating system for the SCO bank, the New Development Bank, et cetera, et cetera. So they don’t necessarily all have to run on mBridge as mBridge exists today, but there is no technical reason that stops them from operating discrete networks that have shared nodes and ledgers where necessary that would enable interoperability between different networks.
Asbed Bedrossian: So would a country sign up its currency in CBETS? And maybe I don’t understand exactly how it works. Does that mean that you become part of the alliance where you are able to trade in your own national currency? And does that affect your ability to be part of SWIFT? because at least in the case of Armenia, we’ve seen this multipolar push-pull where they say, for example, you cannot be part of the Eurasian Economic Union and the European Union.
It’s primarily, in my opinion, political. But in this case, we’re talking about technology, and technology is just software, and things can be done.
Warwick Powell: Well, there are two issues. One is a political question and one is a technical question. There is no technical constraint. All these constraints are political.
There is no reason why. So you’ve got to remember what money is, right? So money, particularly these days, is really entries in ledgers. and when transactions are being undertaken between two parties and they’re being settled between the central banks of the parties involved, what’s happening is that we’re getting deductions from one ledger and additions to another ledger. So that’s all that’s happening.
And what’s needed to ensure systemic integrity is that the deductions do happen and that the additions do happen and that the proportions are agreed. So that’s the issue, right? So when participating banks use platforms like CBETS, they either choose to transact in RMB, the e-RMB, in and of itself, and they can do that basically now. So if a central bank in a country is settling accounts with the PBOC to settle the trades that have been undertaken earlier in the day between companies in these two countries, they can settle those accounts in RMB.
Okay, so that’s something that is likely to happen more of. And we’ll see that grow as central banks around the world increase their holdings of RMB, enter into expanded RMB swap agreements with the People’s Bank of China, but also seek to raise their RMB reserves through the issuance of RMB-denominated bonds.
So they go into the China market, for example, and offer to investors in China, depositors, opportunities to invest in either a commercial or corporate bond or a national government bond, like the Indonesians, the Hungarians have done in recent times. that would enable a Chinese account holder to purchase a bond and earn an interest paid in RMB, but make available to the central bank that has issued the bond the foreign exchange or the RMB in question that will help them settle RMB accounts.
The mBridge system was designed and does support central bank digital currency settlements in multiple currencies, whether it’s RMB or the dirham from UAE or wherever.
Asbed Bedrossian: Does it support the dollar?
Warwick Powell: No, it doesn’t. The dollar’s not part of it. And the dollar chose not to be part of it.
Asbed Bedrossian: Hmm.
Warwick Powell: So the way that mBridge works is that there are six participating central banks at the moment, and each of those runs essentially a node of a computer network.
And that computer network is what maintains this set of books, right, the pluses and the minuses. when parties from one country engage in a transaction with another country the commercial banks reach agreement but ultimately the settlement of that trade takes place at a central bank to central bank basis and there will be agreements on what currency that settlement will be made in and it can actually be a currency in one or the other, you know, participating currencies, but it’s designed to support peer-to -peer transactions and the agreement at the moment would be that the exchange rate is determined on that transaction.
So there is no market for exchange rate determination or price discovery and there’s no permanent mechanism at this point in time for an alternative price discovery mechanism or a price agreement mechanism, exchange rate settlement mechanism. This is all done on a transaction-by-transaction basis. I suspect as time goes by, there will be a range of agreements reached around how exchange rate relativities are to be agreed by the participating central banks. And it may be periodically reviewed quarterly, half yearly, annually.
It may have rules that could mandate adjustments by count by other banks in the event of trade surpluses or trade deficits, for example. I think there will be a lot of lessons drawn out of John Maynard Keynes’ proposals on the Bancor from the early 1940s, as far as adjustment mechanisms are concerned. But they’re the sorts of things that will come as these systems mature. You don’t need these mechanisms in place for these systems to work.
You simply need the ability for the participants to enter into transaction-specific bilateral agreements.
Hovik Manucharyan: Dr. Powell, let’s transition our discussion a little bit. You wrote an article recently, China Emerges as a Great Enabling Power, where you describe China as a curator of open network formation. Can you tell us what that means?
What distinguishes enabling power from a hegemonic power, for instance?
Warwick Powell: An enabling power delivers and makes possible the autonomous development of critical infrastructures, whether they are physical infrastructures, economic transformative capabilities, manufacturing technologies, equipment, etc., as well as the soft infrastructure around human capacity development.
And what we’ve seen in terms of China’s own evolving understanding of itself as an emerging great power is an attempt by China to be one that enables others to pursue developmental objectives in ways that have very, very few strings attached. developmental objectives being achieved through technology upgrading, through accessing technologies that had historically been inaccessible, whether it’s machinery, equipment, energy generating and distribution infrastructure, AI models, open source, et cetera, et cetera.
Hovik Manucharyan: AI models.
Warwick Powell: There’s actually a remark made in the last 48 hours by a senior Chinese figure, not from the government, but from an adjacent organisation who actually spoke about China’s development model. So China talks about there being a Chinese development model or a Chinese modernisation model or modernisation with Chinese characteristics as being an open source development model.
So rather than a development model premised on the idea of extracting from third parties, the idea of an open source development model is that it delivers the capacity for others to also pursue their own development, whether it’s by financing, building infrastructure, building schools, providing training, as well as, of course, making available energy autonomy or energy sovereignty.
And as you mentioned in the new frontier is the AI models that don’t come with the kind of strings that are attached to proprietary models but are essentially open source capabilities that organisations and individuals and countries, you know, states and their departments can literally download and then build their own solutions on top of. So there’s, and that enables others to actually pursue their own trajectories and priorities in ways that are, in effect censorship proof. So once, if you’re thinking, let’s take manufacturing as an example.
Modern manufacturing is actually a complex set of processes that requires the alignment of many different actors with different skills and resources. So that’s supply chain coordination and orchestration. But at the same time, to be able to pull that off, you need to have the right equipment in factories. You need to have people with the right knowledge about that equipment and the materials that that equipment is going to transform.
And importantly, you’re also going to need the energy that will mobilise the machinery, whether it’s transportation machinery or processing machinery. What China has been enabling others to do is build their own factories or erect factories, train people to use the equipment, have access to supply chains upstream for raw materials and intermediate goods, have energy infrastructure that enables others to have electricity to run their factories without being dependent on third parties once built.
So instead of having to run your electricity grid with diesel generators as backup systems, which is of course what tends to happen in many Global South countries these countries now as they implement things like microgrid solutions as well as much larger systems will have their own national electricity capability the issue with being dependent upon imported hydrocarbons for electricity is not only that you don’t have them available to you and you are exposed to global supply shocks, but you’re also exposed to the need to use foreign reserves, foreign exchange reserves, to settle those energy accounts when those foreign exchange reserves could often be better used to do other things, such as buying new machinery.
Because once you’ve got the machinery, you’ve got what you need to do what you want.
Hovik Manucharyan: Yeah.
Warwick Powell: So that’s the way that I understand China’s ongoing evolution as an enabling power rather than an extractive hegemonic power.
Hovik Manucharyan: Dr.
Hovik Manucharyan: Powell, what could invalidate this thesis? Would it be possible that as countries increase their dependency on China through finance, infrastructure, supply chains, could that eventually become another source of coercive power in your opinion?
Warwick Powell: Well, I think there is always a possibility where relationships between states that are cooperative and involve exchanges can also open up a space where those exchange relationships can be coercive. The issue about dealing with all of that, of course, is to have options available. Okay, so a very, you know, trade coercion is a great example of this, right? So I’ll just speak generally rather than pick on any particular cases, but trade is often used as a way of expressing displeasure diplomatically, and that’s also often used as a way of coercing behaviours from others by creating sanctions or what have you.
What we know, though, is that those sorts of measures are becoming increasingly less effective in a world where countries and organisations have more choices. So when you have the ability to transact with more than one counterparty where there are a set of global or multinational rules and procedures that are beyond the reach of single actors to censor, for countries to get themselves or mitigate the risks of coercion, number one, but also it sends a very powerful signal the other way, which is you might express displeasure, we get that, right?
But if you think that expressing displeasure through these so-called coercive measures is going to change behaviour, you’ve got something else coming. It’s not as easy today to coerce behaviours out of others. States. The United States has, of course, historically used all sorts of coercive measures to pursue its foreign policy objectives and to change the behaviour of other states.
But those measures have become less and less effective. And one of the reasons is that states have choices that they can go elsewhere. But there’s also a broader range of institutional environments that provide a certain level of protection, if you will, from these kinds of coercive measures.
Trade is a good example because things like the World Trade Organization and all of the multilateral trade agreements that exist underneath it and bilateral trade agreements for that matter provide the means by which transacting parties are able to continue finding new markets in ways that they can have reasonable confidence in procedures and operational requirements. regardless of what the United States is doing. So the United States can coerce as much as it wants, but if countries know that they have access to these alternative milieu and alternative networks and channels, then the effectiveness of coercive tools diminishes.
And I think one of the things for China, and this is very different, is that China, as big a state as it is and as significant and powerful as it is, to be operating in an environment which has an obvious centre in lots of subordinate peripheries. The development of countries really means that it’s so much harder, even if one wanted to, to throw one’s weight around. And I think the Eurasian continent’s a very good example of that, where you’ve got Russia, of course. You know, Russia is no minnow.
You’re not going to be pushing Russia around. shown that as a country of 90 odd million, it’s not going to get pushed around. Attempts to coerce Iran for four plus decades have actually not succeeded. Another great example of where coercive measures often run into resistance from people in countries and institutions themselves. So, yeah, so that was, so whilst I do think that there is always a risk that the biggest player in a He’s tempted by these.
It’s not as easy anymore.
Asbed Bedrossian: Yeah, the United States and Europe are increasingly supporting transport and infrastructure project across Eurasia. And in the South Caucasus and in Armenia, they’re sponsoring the TRIPP project, the Trump route, which is also part of the larger so-called Middle Corridor, which includes the Trans-Caspian corridor and everything. You have previously argued that transport networks can become complementary, right, rather than mutually exclusive. TRIPP offers an interesting example of that because of the direct American role of
Hovik Manucharyan: Or a test.
Asbed Bedrossian: the project. Pardon me?
Warwick Powell: Thank you.
Hovik Manucharyan: Rather than an example, maybe a test, whether it’s possible.
Asbed Bedrossian: Yeah, that’s right. It’s a test. So in this light, Professor, how does Beijing actually view the TRIPP project and the wider Middle Corridor? Does it see them as complementary to its Belt and Road Initiative or geopolitical competition or as an effort to constrain China’s connectivity with Europe?
Warwick Powell: Well, probably all of the above is the answer.
Asbed Bedrossian: Hm.
Warwick Powell: And a lot of it, how infrastructure gets used is actually the open question. And there’s ways in which infrastructure can be used in ways that are complementary to each other, which support national economic development objectives, which provide countries and businesses with choices and alternatives, or they can be used as choke points. the way to mitigate the risk of infrastructure channels being used as choke points is of course to have access to others um one of the things that you do find with
Asbed Bedrossian: Mm-hmm.
Warwick Powell: Chinese foreign policy is that they tend not to um get themselves too worked up over you know whether the the United States or Europe or others is building infrastructure the key is that there is complementary or at least alternatives available for countries to use and if those alternatives exist then in many ways the coercive potential of certain things diminishes um so uh whilst i obviously don’t have a direct line into the thinking of uh the Chinese leadership uh judging by the ways that they’ve usually responded to developments around the world there tends to be um uh relaxed maybe a little bit of a I don’t think that Beijing gets as anxious about these sorts of infrastructure
Warwick Powell: projects being developed by the West.
Asbed Bedrossian: But let me just say that, you know, they’re big, they’re powerful. So, of course, it tempers the panic. But we’ve had at least two bad examples in 2026. We had Venezuela and we’ve had Panama.
In both cases, the United States was very, let’s say, aggressive in its posture towards China, saying that these countries are somehow, they’re somehow offending trade with the United States because they’re doing too much trade with China.
Warwick Powell: Well, I think it speaks to firstly American anxieties
Asbed Bedrossian: And of course, you know, I don’t know what that example means for China.
Warwick Powell: about the presence of China in its own backyard. That’s the first thing. Second thing, it also speaks to a certain style of American foreign policy conduct which resorts to assorted forms of violent interventions, you know, such as kidnapping a national leader and essentially running a regime change operation. This has been par for the course for many decades, of course.
Look, the Venezuelan example is a very interesting one because it continues to export oil to China anyway. What’s the main difference? Well, the main difference is that the payment is now being interceded by the United States Treasury. But even then, I think that people make up too much of the issue of Venezuelan oil in China.
China’s energy dependencies are actually incredibly distributed and diverse. And it has been far better prepared for the kind of energy disruptions that we’ve witnessed in the last six months than I think many people thought China was. So that’s the Venezuela story. The Panama story is actually a fascinating case because we’ve had two, Well, the port infrastructure was operated by a Chinese registered company, obviously got pushed out.
But now, of course, China’s responded to that through what are essentially administrative measures that has put major shipping operators on notice. You can do business with Panama if you want, but if you then think that those ships are going to come into the Chinese ports, you’ve got another thing coming. So in a sense, China’s responded to that particular circumstance through administrative means. The last thing I’ll say about Panama, I wrote a paper on this last year or the year before now.
Time flies. I forget when exactly. But the Panama question is no longer as significant as perhaps it once was, in part because of the development of the port in Peru, the Chancay port, in part because that port can also handle the new generation of massive container ships that the Panama Canal can’t. And the Panama Canal actually can’t ever handle ships of that size, largely for water and environmental reasons.
And so Panama itself is of less economic significance than it historically once was. And so keeping these things in context, yeah, sure, you know, the Americans were coercive in both of those examples and the Chinese state has responded in ways that perhaps, you know, didn’t accord with the beat the chest mentality that many people have, but largely through administrative measures. And currently on the Venezuelan issue, you know, I don’t think China’s ever going to militarily intervene
Asbed Bedrossian: I can say that there’s hardly any coverage of China’s responses in mainstream media
Warwick Powell: in those sorts of circumstances.
Asbed Bedrossian: in the United States.
Warwick Powell: No. Well, of course, because the mainstream media needs to paint a certain story. One is that the Venezuelan operation was successful from America’s point of view, and I think it’s fair to say that from the United States’ point of view, from Washington’s point of view, that operation has been successful. In terms of the Panama situation, I think the situation is far messier.
Washington wanted to bring Panama and the canal back under its remit indirectly, but Panama itself has had a price to pay. And in fact, the operations of that canal has come under incredible commercial pressure as well. Look, there are real issues with that canal. You know, it’s a relatively old piece of infrastructure now, as we know.
It’s constrained environmentally. and so I think that its economic significance going forward over the next century, say, is going to diminish and there’s not much it can actually do about that. And it is because there are alternatives and it’s because it’s actually not really fit for purpose just by scale, really.
Hovik Manucharyan: Dr. Powell, let’s switch gears a little bit more. Now, the last few weeks have been quite interesting in the world of technology and AI. There have been calls to slow down the development of AI.
There have been public resignations of employees who warned that there’s a 10% chance that AI will destroy all of humanity. And rather than responding…
Asbed Bedrossian: Hovik, can I add one thing to that? There’s also conversation about whether AI could become President of the United States. So there’s all kinds of conversation about AI.
Hovik Manucharyan: Well, but what’s interesting is that now leaders of frontier AI companies are lining up. They’re not sort of denying this. Anthropic CEO Dario Amodei wrote a very long essay, We Must Pace the Frontier, essentially asking almost, if you look at it one way, and arguing that model capabilities may be advancing faster than safety systems, and there needs to be some kind of governance. Now, critics to this see some kind of safety debate, basically, see there’s some ulterior motives.
For instance, established AI firms could use safety as a guise, to essentially lock in barriers to new entrants. Now, the response by United States and China is also interesting. Trump administration, or Trump basically himself said they’re not going to regulate anything, they’re going to let AI go forward. And China responded to Amodei’s remarks saying that it’s a Cold War trick, fear-mongering, hampers global governance.
But at the same time, China urged all parties to promote open and inclusive AI development for the good of all humanity. this places this AI debate squarely in the middle of, I think, the struggle over power, over industrial development, and so forth. Where do you see this debate going? Do you see frontier AI development as a genuine systemic threat, or is the pacing debate essentially shaped by commercial and geopolitical interests?
Warwick Powell: Well, all of those things can actually be true at the same time, okay? So, on the safety question, I think that it would be irresponsible to take the attitude that there are no risks, that the pace of AI development and its capacity to operate autonomously in a self -learning mode is real. And we’ve seen some recent examples of that. And therefore, anybody, whether they’re individuals or organisations or nation states, would be irresponsible to just dismiss concerns about AI getting out of hand and running faster than humans can keep up with, simply because there are other agendas at play.
There are also, I think, some very serious commercial agendas at play as well. the American AI business model was premised on the idea that the small number of American AI firms with proprietary technologies would be able to sustain very high walls and very wide and deep commercial moats that nobody elsewhere in the world could challenge them on and by virtue of that were in a position to generate economic rents, sustain revenue levels that delivered substantial above average profit rates. That was the business model.
That business model collapsed when DeepSeek, at the beginning of last year, released its first open source model And the collapse of that American AI proprietary business model was consolidated over the course of the last 18 months as other Chinese AI labs released better and better open source models.
So the proprietary American AI companies are facing some very serious commercial and financial headwinds. circumstances, it is little wonder that the principals of these firms will be looking for ways in which they could gain protective cover in one shape or another to rebuild some moats or build some new moats that would deliver to them protective cover and new revenue security that delivers above average profits. In other words, politically sanctioned economic rents. There is no better cover for that than the question of safety and national security. So it’s hardly surprising that we have these firms pursuing this.
But this intersects with, I think, some broader issues. Clearly, there is a view among certain sections of the American political and technological elite, that artificial intelligence is a zero-sum race and that he who wins AI wins. And because of that, there is going to be pushback from the American political elite to begin with. And we’ve seen that in the way President Trump’s responded.
But we’re also going to see a high degree of global cynicism about the genuineness of American AI techno-principals advocating things that would deliver to them commercial advantage. Now, none of that helps in tackling the issues of substance. It creates political imperatives and it also creates or consolidates and exacerbates commercial concerns.
Now, sitting in behind all of this is the intersection of corporate technological interests, geopolitical interests, and a millenarian ethos that we find amongst certain proponents of American AI, which really sees American AI in particular, and AI broadly speaking, as existential in nature. for humanity, but existential in terms of civilizational success. And this relates to the Dark Enlightenment philosophies that are part of certain sections of the American AI community.
And so you put all of this story together, these elements, these core elements together, and I think we can say that there are some genuine issues and it would be irresponsible to just brush them aside. But the ability of companies and governments to deal with them at this current juncture is incredibly difficult because of the intersection of these other forces at work. So in the short term, my own sense is that the United States in particular will actually find it very difficult to address these issues, the issues of existential risk and risk to humanity generally, on its merits, in a way that is persuasive to bring others on board.
This is a very difficult time because clearly artificial intelligence cannot be constrained by national borders . And because of that, it is imperative, in fact, that people who take these issues seriously across national borders are able to coordinate a way of thinking about these issues and to put in place mechanisms by which they can govern them. In the past, and I’ll just end on this one last observation, when there were other technologies that had existential risk on the table, namely nuclear weapons. At the time, the Soviets and the Americans were the players.
They were able to ultimately reach certain kinds of detente around how these technologies were to be able to be used and developed, but that was in a different era. We’ve now got levels of distrust underpinned by this millenarian apocalyptic zealotry from particularly sections of the American technological and political elite that I think is making the environment, it’s polluting the environment in ways that is going to be very difficult for the foreseeable future to overcome.
Now, the only things that I think we can look forward to in the short term is that there are, at least at a superficial level, some agreements around some consultative mechanisms that at least create the platforms through which regular and ongoing discussions can begin to take place.
Asbed Bedrossian: Professor, a couple of questions about the full stack of AI and also AI sovereignty. AI spans more than just large language models. It spans the entire stack of technology, from semiconductors to foundries to energy to infrastructure. For example, even Armenia is thinking about how it’s going to manage its nuclear power plants to have data centers and such.
But beyond competing for the AI race, nations not called the United States and China have a fairly large problem, many of them in Southeast Asia. And you discussed this in your interview with John Pang recently. They have to navigate competing ecosystems to avoid technological dependency. I was calling it technological destitution almost because the cost of these AI systems is pushing certain countries into stratifications, the haves and the have nots.
So there’s a new set of problems that these countries are going to face unless they have access to this new platform. are the real geopolitical bottlenecks in the AI stack currently? Are they chips, compute energy? Where do you see this going?
Warwick Powell: Well, I actually think energy is your foundational bottleneck. And because AI, interestingly, has brought to the front a genuine understanding that information is actually an energetic artifact.
And in very measurable ways, measurable in terms of energy consumed for tokens produced. right so we now have an environment where we actually have a very granular understanding of of how energetic information actually is so my suggestion is is that our first order of business if you will particularly from the point of view of risk and lock-in and all of that is addressing this question of energy addressing this question of energy opens up a Pandora’s box we touched on of energy sovereignty, but we also need to be thinking about questions of energy haves and energy have-nots. This is a critical issue for societies at a domestic level.
So you might solve a national question around energy, but if you don’t at the same time ensure that in solving your first problem you don’t create an access problem for your society at large, I think you will end up creating more problems than you have solved. And we see that in the United States where the expansion of AI data centres, these big hyperscale data centres and the impacts that’s having on the electricity systems and electricity markets is now leaving a legacy of electricity disadvantage, either in terms of cost itself or worse than that in terms of actual constraints on availability. So I think energy is at the substrate.
The second issue will ultimately be chips, right, because that stuff has to get made, right, and it’s made through complex processes and complex technologies. And being able to ensure that you have available to you the hardware, the tools, is going to be absolutely critical. and so from a national development point of view, they would be critical issues. Though having said that, I think one of the things that policymakers need to be mindful of is the rapid evolution of distributed artificial intelligence capabilities.
So far, AI has seen the rapid expansion of big data centres Big data centers will continue to play incredibly important roles in how AI works, but we are seeing the developments in a whole range of technologies that is going to enable a lot more computation and data storage to take place at the edges of networks, whether it’s in small devices, not just devices like phones, which are actually pretty powerful for their size, but in much smaller devices, little IoT devices. sensors, et cetera, et cetera.
And the ability to deliver this kind of capability at the edges will make a big difference in both good and potentially risky and dangerous ways.
Going back to that issue of existential risks, one of the concerns or the foundational concern is that AI within an agentic environment, multiple agents, multiple devices operating autonomously but nonetheless coordinating develops a quote a life of its own and the more we can push compute and AI to the edges the more powerful this network of and and the greater in number this network of AI enabled agents become which obviously exacerbates certain kinds of risks but at the same time creates a certain set of opportunities So this issue of, so that’s your two sort of core pieces, I think.
In terms of applications, the fact that there are open source models now, in a sense, takes that question of risk out other than being able to have the hardware to run it and the electricity to operationalise it.
The big opportunities in all of this incidentally, I mean, we can talk a lot about risk, but I think the bigger opportunities for countries is also going to be in how they make use of these foundational capabilities and have sufficiently trained people. and then to rapidly develop applications where these core technologies find their way into the daily lives of social and economic systems so that these particular systems begin to benefit from better information, quicker information, more effective coordination, et cetera, et cetera.
You know, so never lose sight of the opportunities whilst at the same time, I think, you know, being incredibly clear-eyed about where many of these risks are.
Asbed Bedrossian: That’s a very interesting thing that you mentioned, because we were talking about AI sovereignty, national sovereignty, now, these couple of data centers that Armenia is going to be setting up, the government has been claiming that it enhances Armenia’s national security. Meanwhile, we have seen, for example, having something like Taiwan Semi in Taiwan made it so that the superpowers always approach it with kid gloves, right? with kid gloves, they don’t want to lose Taiwan Semi. China doesn’t want to lose Taiwan Semi, but wants Taiwan.
What is exactly the component that’s going to enhance a country’s, let’s say, national security or sovereignty or technology sovereignty? Is it just a data center or is it going to be? Yeah, you tell us.
Hovik Manucharyan: Is there any future in democratization of chip production, for instance? What does the future look like for that?
Asbed Bedrossian: Thank you.
Warwick Powell: Well, it’s a technically and scientifically complicated set of activities. The idea of being able to produce these things anywhere, anytime, I think is certainly not going to be the case in my lifetime. um but um but but i think one of the keys to be focused on is again options choices um rather than exposed to one or another so that’s that’s an important part of any consideration this data center question i think is uh it is a fascinating one because like many things in the world, things are often double-edged swords, right?
Having data centres physically located within one’s country and governable by the laws of that country and arguably able to be disconnected from external sources, you know, by turning off data cables or what have you, in a sense gives a country a capacity to say, well, at least we have a protected ecosystem. But at the same time, big data centres are also targets, not only for cyber security attacks, but we’ve seen in West Asia data centres being targeted by missiles. Okay. So double-edged sword.
Asbed Bedrossian: Absolutely.
Warwick Powell: The equipment that’s run in them, of course, in many, often the case from other parts of the world, because not everywhere it manufactures all the things that you need. So there are issues there. Keeping them cool and running them is an energy question. The upside is that energy sovereignty is a lot easier to solve than sovereignty on other things.
And I’d argue, in fact, that for things where sovereignty isn’t actually very easy to solve, such as the manufacture of microprocessors, the most important thing for small countries that don’t manufacture these sorts of things is to ensure that the world has many options available and to avail themselves of as many of those options as possible. The difficulty you’ve got is that the United States, and I know this in other contexts, is trying to strike agreements on things like data centres with countries that essentially has poison pills that prohibits these countries from having agreements with others without the approval of the Americans.
So there is an attempt by the Americans to lock countries into ecosystems, whereas I think from the point of view of questions of national sovereignty, the resistance must be to not be locked into any ecosystem but to ensure that you have the availability of choice to you.
The upside here is that the technological developments on distributed capabilities does open up potential for a lot more AI to happen at the edges where the computational demands on things like the latest microprocessors is actually not as high so you can actually use not necessarily the latest of of equipment or the best of the equipment available to nonetheless do the things that you want to be able to do. And that’s actually going to be important as well for developing countries and others.
So I think that from a sovereignty perspective, even though there are the risks that I touched on earlier, strategic thinking needs to go into enabling a more distributed network within nations themselves. And the last point to really be thinking about now is what I call the information integrity architecture.
How is it that stakeholders within a society, governments, individuals, households, companies, organisations or what have you, can participate in an information ecosystem where there can be a high level of confidence that the information available and that’s being produced and processed and stored within this ecosystem actually has the integrity that participants expect and need.
That’s going to be the critical question and it’s a political question as much as it is a technical question because the downside risk of the rapid development of information systems that we’ve experienced in our lifetimes is that we can produce more information quicker than we’ve ever been able to do.
And yet, of that information that we are producing quicker than we’ve ever done before, I would suggest that a growing proportion of that is noise at best and, in fact, quite destructive and misleading nonsense at worst. we’re now confronting not only the physical realities of information systems, whether it’s energy, whether it’s chip production or what have you, we’re actually confronting a much broader social, political, cultural question around information integrity. And those debates have not yet opened up properly either.
Hovik Manucharyan: Right.
Hovik Manucharyan: Dr. Powell, I want to go back to something you said earlier because I think it’s very relevant to our Armenian audiences. You mentioned that the United States has a tendency essentially to lock in countries exclusively.
So if the United States, for instance, has approved the use of NVIDIA GPUs for Armenia, the Armenian government currently is not being very transparent in what it has exactly signed with the United States on this and a number of other deals so I’m wondering if such agreements also exist with the government of Armenia where Armenia, having purchased NVIDIA GPUs, doesn’t have the option of now going outside of that are there examples you said other contexts Are there examples of where you’ve seen this happen?
Warwick Powell: Well, you mentioned my discussion recently with John Pang from Malaysia, and the Malaysian agreements is a classic example where these concerns have now come to the fore. And there are some other agreements that have been tacked onto the tariff agreements and the trade agreements that the Trump administration pursued in the last 12 months, which seeks to lock countries into particular technological ecosystems. But the Malaysian case is a reasonably well-known case, and certainly for me here in the Asia-Pacific of a country.
And, of course, you’ve got to remember that Malaysia has, over the last couple of decades, been actively developing a microprocessor industry to manufacture and support the global supply chains of chips. And it has also been developing a regional data centre industry as well. to literally be the data crossroads for North Asia, the Pacific, the Indian Ocean and things like that. So Malaysia has a very particular footprint and an interest in what goes on in this space.
And the agreements that were agreed to, as I understand them, does have so-called poison pills in them, which is that the Americans have to agree to the inclusion of other technologies within the infrastructure. And that has become very controversial in Malaysia. Now, there are stories, and I don’t have them validated, but we’re starting to see more stories about how different ecosystems are now operating inside the same buildings on different floors.
But again, those agreements tend to focus on the hyperscale data centres, but they don’t deal with that distributed network of sensors and et cetera, et cetera, where I think we’re going to have a whole new batch of issues, both risks and opportunities. So from a small country’s point of view, such as Armenia, I would think that you need to actually have a look I wouldn’t be surprised if these agreements have these poison pill clauses in them, which gives the United States a veto in effect. And I wouldn’t be surprised if those vetoes relate to hardware and software lock -ins.
So there are some real risks for small countries who commit to these sorts of things because you would basically be denying yourself the kinds of choices and options that I mentioned earlier, which are actually central to a small nation’s ability to navigate this much more fluid and complex world.
Hovik Manucharyan: Very interesting.
Asbed Bedrossian: Professor Powell, it’s always a pleasure to talk with you. You raise so many questions that we just have to keep thinking and talking about these issues. We appreciate the time and the insight you give us.
Warwick Powell: Look, it’s been a pleasure. I’m sorry that I couldn’t give more definitive answers because I think right now we’re confronted more with the questions than we are with the easy answers.
Hovik Manucharyan: Yes.
Warwick Powell: But these are the things that are testing us all and for small countries
Asbed Bedrossian: Yeah.
Warwick Powell: in particular. And I come from a small country. These are also the problems that are plaguing communities and enterprises and regulators. So maybe through more dialogue and exchange, we can find some common answers to how we protect ourselves and also protect humanity.
Asbed Bedrossian: All right, well, that’s our show today. This episode was recorded on September 17 in our time zone and September 18 in Professor Warwick’s zone, 2026. So, Professor Warwick Powell is an adjunct professor at Queensland University of Technology in Brisbane, Australia, that’s where he was for this interview, and a senior fellow at Taihe Institute in Beijing. is the author of China, Trust and Digital Supply Chains, Dynamics of a Zero Trust World, and other books. His work focuses on digital innovation, technologies, and international political economy.
For more information, go to our show notes. will be many links. And also that YouTube interview with Warwick. We’ll put the links all in the show notes. podcasts.groong.org slash episode number.
Hovik Manucharyan: us know how we’re doing, folks, in the comments. Subscribe, like, share, and all the other good stuff. Thank you for watching us. My name is Hovik Manucharyan.
Asbed Bedrossian: And I’m Asbed Bedrossian. We’ll talk to you soon. Bye-bye.
Hovik Manucharyan: Until next time, bye bye.