Moody's analysts discuss AI's trade-driven growth in Asia, market corrections, and social labor risks
Summary
Moody's analysts Gaurav Gorev, John Bromhead, and Stefan Angrick analyze AI's economic impact on Asia, arguing that trade in semiconductors, not domestic data center build-out, is the primary growth driver for economies like South Korea and Taiwan. They highlight significant market corrections in Asian tech indices and private valuations, noting that wealth effects are concentrated among wealthy households while private consumption stagnates. The speakers warn of social friction from record bonuses in AI-linked sectors and labor market tensions as AI adoption displaces lower-end knowledge work. They also discuss financing risks in the AI sector, where self-reinforcing debt loops could reverse if demand disappoints, and note the vulnerability of service-export economies like India and the Philippines to AI-driven job displacement.
Episode sections
- 0:20 – 3:34
Introduction and AI Economic Channels
Gaurav Gorev introduces the episode's focus on Asia and AI, outlining three channels of economic impact: physical data center investment, AI adoption productivity gains, and financial market wealth effects.
- 3:34 – 8:02
Data Center Build-out and Regulatory Pushback
John Bromhead discusses Australia's data center construction boom, noting labor shortages and community opposition leading to regulatory scrutiny, while highlighting South Korea's strong Q3 GDP growth driven by chip exports.
- 8:02 – 11:59
Trade vs. Investment in AI Growth
Stefan Angrick argues that AI's primary economic driver in Asia is trade (chip exports) rather than domestic data center investment, citing massive export growth in South Korea, Singapore, and Taiwan.
- 11:59 – 15:29
Top Beneficiary Economies
The speakers identify Taiwan, Singapore, Malaysia, and South Korea as the top economies benefiting from AI-related trade, while noting China's diversified export base and Japan's offshore production model limit their direct statistical impact.
- 15:29 – 17:06
Uneven Wealth Distribution and Consumption Stagnation
John Bromhead explains that despite strong GDP in South Korea, private consumption has flatlined, with wealth effects concentrated among wealthy households and institutions rather than driving broad consumer spending.
- 17:06 – 20:04
Record Bonuses and Social Friction
Stefan Angrick notes that record bonuses for workers in AI-linked sectors in South Korea are creating social tension and highlighting the disparity between winners and losers in the economy.
- 20:04 – 23:03
Equity Market Corrections and Rotation
John Bromhead highlights significant declines in Asian tech indices (COSPI, Nikkei, Taiwan) and private valuations (Anthropic), contrasting this with resilience in India and a market rotation away from AI capex exposure.
- 23:03 – 30:25
US Political Risks and Future Pipeline
The speakers discuss how US midterm ballot measures against data centers and Democratic wins in key states like Texas may slow future US investment, potentially shifting demand to Asia or improving unit economics for existing providers.
- 30:25 – 33:15
AI Financing Risks and Debt Loops
Speaker_0 expresses concern that tightening monetary policy and rising debt costs could turn the AI investment boom into a broader financial market dislocation. Speaker_3 explains the self-reinforcing financing cycle where data center contracts serve as collateral for more lending, creating a virtuous cycle that could reverse into a vicious one if demand disappoints.
- 33:15 – 35:36
Valuation Volatility from Private to Public Markets
Speaker_3 highlights that US tech profits are partly driven by mark-to-market gains on investments in frontier labs like Anthropic and OpenAI. He warns that as these private companies go public, daily valuation changes could introduce significant volatility compared to the current creative accounting environment.
- 35:36 – 38:02
Social and Labor Market Impacts in Asia
John Bromhead notes that AI adoption in Asia is creating labor market tensions, with young people struggling to find jobs as firms replace lower-end knowledge work with AI agents. Speaker_0 adds that middle-income Asian economies lack the fiscal capacity to implement social safety nets like UBI, raising risks of social instability.
- 38:02 – 39:35
Risks to Service Export Economies
Bromhead expresses concern for South Asian economies and the Philippines that rely on service and IT exports, as AI threatens the traditional export-led growth model. Speaker_0 acknowledges this is a medium-term risk, noting that while India's BPO sector hasn't yet been significantly impacted, the vulnerability remains.
- 39:35 – 40:22
Asymmetry of AI Economic Effects
Speaker_0 describes an asymmetry where AI benefits are currently concentrated in chip exports and data center build-outs, while negative effects like job losses could spill over rapidly to other sectors. This contrasts with the broad-based benefits typically seen from historical technologies.
- 40:22 – 44:46
UN Job Shortfall Projections
Speaker_3 cites UN Development Programme research projecting 1.2 billion young people entering the workforce in emerging economies over the next 10-15 years, against only 400 million new jobs, highlighting a structural shortfall exacerbated by AI.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| AppleAAPL | equity | bullish | it's a little bit similar to what we saw in the US with like Apple outperforming the other hyperscalers |
| NVIDIANVDA | equity | mixed | And and the latest episode with uh NVIDIA doesn't help, does it? Now that's also taken a fair bit of shine off NVIDIA stocks. |
| AlphabetGOOGL | equity | neutral | So a Google or an Anthropic or any of these kind of data center builders sign a multi billion dollar contract, compute contract. |
| AmazonAMZN | equity | neutral | And what happens over time is that those valuations go up and Google and Amazon and whatever get to revalue their positions and basically report it as profits. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
South Korea's Q3 GDP came in at 1% quarter-on-quarter.
Unidentified speaker (UNKNOWN) ·5:18
How to verify: Bank of Korea GDP release for Q3 2026. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
If semiconductor exports and chip facility investment were flatlined to long-term trends, South Korea's Q3 GDP would have been 0.2%.
John Bromhead ·5:39
How to verify: Moody's Analytics research or Bank of Korea statistical breakdown of GDP components. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Half or slightly more than half of South Korea's current economic growth is coming from the data center story and chip investment.
John Bromhead ·5:48
How to verify: Bank of Korea research paper cited by the speaker. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
South Korean exports are up more than 90% year-over-year.
Unidentified speaker (speaker_2) ·9:17
How to verify: Statistics Korea (KOSTAT) export data for the relevant period. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Singapore exports are up north of 60% year-over-year.
Unidentified speaker (speaker_2) ·9:37
How to verify: Singapore Department of Statistics export data. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Taiwan's GDP is growing at double-digit rates this year.
Unidentified speaker (speaker_2) ·11:15
How to verify: National Development Council of Taiwan GDP reports. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Singapore and Malaysia GDP is growing north of 5%.
Unidentified speaker (speaker_2) ·11:15
How to verify: National statistics agencies of Singapore and Malaysia. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Private consumption in South Korea flatlined in Q3, not happening in nearly four years.
Unidentified speaker (UNKNOWN) ·15:52
How to verify: Bank of Korea consumer spending data. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
COSPI is down over 20% since the start of October.
John Bromhead ·20:27
How to verify: Korea Exchange (KRX) COSPI index performance data. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
47 out of 48 US ballot measures on data centers voted against more investment.
John Bromhead ·23:45
How to verify: US state election results and ballot measure records from the 2026 midterms. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Financing conditions are tightening, leading to higher debt costs and risk appetite tightening, which could cause AI stock corrections to spill over into broader asset classes.
Unidentified speaker (speaker_0) ·31:05
How to verify: Check Federal Reserve meeting minutes or H.15 reports for monetary policy tightening trends; monitor VIX and cross-asset correlation data for market dislocation signals. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Many data center deals are funded by debt backed by the AI ecosystem itself, where user contracts serve as collateral for financing chip purchases and construction.
Unidentified speaker (speaker_3) ·31:45
How to verify: Review prospectuses or financing documents for major data center REITs or cloud providers to identify debt structures and collateral types. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
CVS (Credit Default Swaps) for neo-cloud providers, particularly CoreWeave, have blown out, indicating market stress.
Unidentified speaker (speaker_3) ·32:53
How to verify: Check CDS spread data for CoreWeave on financial data platforms like Bloomberg or Refinitiv. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Big tech companies report profits from revaluing their positions in frontier labs like Anthropic and OpenAI as valuations rise.
Unidentified speaker (UNKNOWN) ·33:32
How to verify: Analyze quarterly earnings reports (10-Q/10-K) from Google and Amazon for 'equity method earnings' or 'unrealized gains' related to investments in AI startups. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Primary sourceAnthropic is expected to go public next week.
Unidentified speaker (speaker_3) ·34:12
How to verify: Check SEC filings for S-1 registration statements or news releases from Anthropic regarding IPO timelines. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Primary sourceThe UN Development Programme projected 1.2 billion young people entering the workforce in emerging economies over the next 10-15 years, with only 400 million jobs projected.
Unidentified speaker (speaker_3) ·40:34
How to verify: Search UNDP publications or press releases from November 2026 for the specific demographic and employment projections. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
India's business process outsourcing sector has not yet been significantly affected by AI adoption.
Unidentified speaker (speaker_0) ·39:11
How to verify: Review industry reports from NASSCOM or IT sector associations in India regarding employment trends and AI adoption rates in BPO. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
AI is replacing lower-end knowledge work in firms, creating labor market tensions for young people.
John Bromhead ·36:26
How to verify: Review labor market statistics and anecdotal evidence from Asian labor bureaus or HR technology firms tracking AI automation in entry-level roles. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
The AI boom is creating distortions and the market is not yet in equilibrium, requiring a correction to price AI fairly.
John Bromhead ·41:55
How to verify: Analyze valuation multiples of AI-focused companies compared to historical tech bubbles or fundamental cash flow analyses to assess pricing efficiency. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Technology is generally beneficial for human development and economic growth, but the net impact of AI is uncertain.
Unidentified speaker (speaker_0) ·42:45
How to verify: Review historical economic data on technology adoption vs. GDP growth and current economic forecasts for AI's contribution to productivity. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Do your own research
- Verify South Korea's Q3 GDP and semiconductor export data via Statistics Korea (KOSTAT) and Ministry of Trade, Industry and Energy reports.
- Check Kospi, Nikkei, and Taiwan Weighted Index performance data from Bloomberg or local exchange filings for October 2026.
- Review US Census Bureau and EIA data on data center construction permits and ballot measure results from the 2026 midterm elections.
- Analyze CoreWeave and other neo-cloud providers' credit default swap (CDS) spreads on platforms like Markit or Bloomberg to assess financing stress.
- Examine US tech companies' 10-K and 10-Q filings to isolate mark-to-market gains from equity stakes in frontier labs versus core operational earnings.
- Consult UN Development Programme (UNDP) reports on labor force projections and job creation forecasts for emerging economies.
- Assess India and Philippines BPO sector revenue trends and AI adoption rates via company earnings calls and industry reports from NASSCOM or PHILBIN.
- Monitor South Korean household consumption data and bonus payout statistics from Statistics Korea and major bank reports to gauge social friction risks.
Full transcriptexpand
- 0:20
- Unidentified speaker (speaker_0): Hello, and welcome to GEU, the macro podcast that delivers valuable insights into the global economy. I I'm am Gorav Gorev Gangu. Ganguli, head of international economics at Moody's Analytics, and we have a bit of a Asia focus for you today. Asia and AI. Lots of things going on with Asia and AI. Um, no one better placed to talk
- 0:40
- Unidentified speaker (speaker_0): about it than the two people I have together with me today. Stefan Angrick, who many of you are familiar with, and John Bromhead from our office in Sydney, who's who is, amongst other things, our financial market sector. markets expert in Asia Pacific. And John, some of you may be familiar, has actually been on this show before. So welcome back, John. How are you doing here?
- 0:59
- John Bromhead: Yeah, good to be here. Looking forward to our conversation.
- 1:02
- Unidentified speaker (speaker_0): Me too. Me too, John. But first, how is Sydney these days?
- 1:07
- Unidentified speaker (UNKNOWN): No,
- 1:07
- John Bromhead: no, Sydney, uh wonderful time of year. It's um uh just coming into spring and uh all the flowers are out so yeah great time to be in Sydney.
- 1:18
- Unidentified speaker (UNKNOWN): You're
- 1:18
- Unidentified speaker (speaker_0): gonna paint us a rosy picture of AI as a result no doubt. Well Stefan um Tokyo Japan Japanese exports the Japanese economy benefiting a bit from the AI boom I guess but we'll talk a bit about that. We'll debate that. That's So maybe let me start there. If we want to talk about Asia and AI I was trying to think about a good
- 1:38
- Unidentified speaker (speaker_0): way of of beginning the conversation. Many people might be focused immediately on the tremendous stock market gains this year so far. Though no some of that some of that's been coming off over the last few months, but nonetheless pretty impressive stock market gains in some parts of Asia Pacific, all centered around the AI boom. But given that we're economists, maybe we should start the conversation
- 1:58
- Unidentified speaker (speaker_0): conversation somewhere else. There is clearly a big global build build-out out of AI.
- 2:03
- Unidentified speaker (speaker_0): That's the general impression people have. But when you dig beneath the surface, it's not really that truly global in terms of the build-out of AI. So maybe we should pause to just decipher this a bit and talk about what is actually happening across Asia. Because I see three parts to this. One is the actual physical investment that goes into
- 2:23
- Unidentified speaker (speaker_0): data centers. That's one way in which the AI build out benefits economic growth. The other is through the adoption of AI, which leads to productivity gains. Now some people say that's a long run story, not a short run story, but it's interesting to hear what exactly is happening in Asia when it comes to the adoption of AI and how we think that might be
- 2:43
- Unidentified speaker (speaker_0): benefiting economic growth in the region.
- 2:46
- Unidentified speaker (UNKNOWN): um,
- 2:46
- Unidentified speaker (speaker_0): but Also possibly some negative effects in terms of job losses. And the third is the financial aspect of things, how AI is benefiting companies in the region, it's leading to an increase in AI valuation. And investors are participating, of course, in this, which leads to an increase in investor wealth, which in turn can spill over into
- 3:06
- Unidentified speaker (speaker_0): the economy and create um beneficial effects through increased spending, people feeling flush with cash and going out and spending, basically. So I see these three. channels and maybe I've missed something. So let me just open up to you gentlemen here and get your thoughts. On the first one, how is Asia Pacific participating in the economic aspect of things and where is Asia,
- 3:26
- Unidentified speaker (speaker_0): which countries in Asia Pacific are actually building out data centers? I'm gonna start with you, John, because I know that things are happening in Australia.
- 3:34
- Unidentified speaker (UNKNOWN): Well
- 3:34
- John Bromhead: I mean it it it's interesting because um one of my friends is a um consultant for a data stre data center construction firm and they essentially just can't get enough people um in the door like it's it's going mad here. Um that said there's it's also been quite
- 3:54
- John Bromhead: an interesting period of time because our biggest uh data center operator uh recently is its share price halved of the kind over of last I don't know three months or so, given massive community uh opposition to one of its flagship data centers in Victoria. Um so
- 4:14
- John Bromhead: some of the themes seem to be kind of resonating globally
- 4:17
- John Bromhead: too. I mean, this this this this pushback on data centers, and maybe we'll talk a little bit about the US midterms, but um
- 4:25
- Unidentified speaker (UNKNOWN): it
- 4:26
- John Bromhead: it's kind of like
- 4:28
- John Bromhead: it was the Wild West for a little bit on data center construction. And I think now it's it's running up against some regulatory barriers.
- 4:36
- Unidentified speaker (speaker_0): Do you have a sense of how much data center build-out is actually impacting GDP growth investment spending in Australia?
- 4:46
- John Bromhead: Uh It's a reasonably big chunk of investment, but still reasonably.
- 4:51
- John Bromhead: small. Um
- 4:52
- Unidentified speaker (UNKNOWN): what
- 4:53
- John Bromhead: is interesting though, uh you know, also I mean, maybe shifting tax slightly to South Korea, which is obviously at the center of um
- 5:03
- Unidentified speaker (UNKNOWN): of a
- 5:04
- John Bromhead: lot of this AI story, but um they recently had a a Q3 GDP print. And um I did run the numbers. Um the Q3 GDP came in at 1%. so very strong. That's quarter on quarter. Um
- 5:18
- Unidentified speaker (UNKNOWN): if
- 5:18
- John Bromhead: you flatline
- 5:20
- John Bromhead: exports of semiconductors, just if you take the growth rate that they would grow, you know, their long-term growth rate rather
- 5:27
- Unidentified speaker (UNKNOWN): than what
- 5:28
- John Bromhead: the recent trend has been. And if you take the investment into um chip facilities and fabrications, and you also flatline
- 5:37
- Unidentified speaker (UNKNOWN): that, um
- 5:39
- John Bromhead: then
- 5:39
- Unidentified speaker (UNKNOWN): it
- 5:39
- John Bromhead: would have been 0.2% rather than the 1% we got. Um and the and the Bank of Korea actually did some recent research which says that you know
- 5:48
- John Bromhead: uh half or a little bit more than half of Korean um economic growth at the moment is coming from uh the the the data center story and and chip investment. So um for Australia I'd say a little bit less, but for for for places like Korea that are at the epicenter of this stuff, it's you I mean you can't really understate
- 6:08
- John Bromhead: the um importance um to the economic story.
- 6:16
- Unidentified speaker (speaker_0): Which is actually selling the goods that are needed to build our data centers. Australia is not really doing that though, is it it's it's building out data centers and is building them as far as I understand fairly rapidly. And now you're see you're seeing you're sending sensing some pushback here.
- 6:32
- John Bromhead: I think part of it is that there are um there are regulatory issues at play here. So I think that the Australian government's keen to be able to place some regulations on how some of the models work and and uh and the output that they provide. But I think that they need to have a physical
- 6:52
- John Bromhead: um kind of presence in order to do that. So um so that's kind of one reason and and um I think the other reason is just that like there is huge, huge demand. I mean Australia has been, I think on a per capita basis, one of the biggest users of AI um and and remains that way. So
- 7:12
- John Bromhead: um very high levels of adoption
- 7:16
- John Bromhead: and
- 7:17
- John Bromhead: huge per capita spending as well. Even h at a household level, a lot more people here are paying for the Claudes and OpenAI subscriptions. So there's huge in inference demand as well.
- 7:29
- John Bromhead: And the yeah, I I think and and we'll talk about it. I mean, the question is is is where you build these things and whether the there's the power infrastructure to feed them. Um, and that's still a very much an open question here.
- 7:43
- Unidentified speaker (UNKNOWN): And
- 7:43
- Unidentified speaker (speaker_0): that that's a question you raised another interesting point, which is the adoption of AI. So let's but but let's bark that yet again. I'll come back to that in a bit. I want to look at other countries in Asia to understand which countries are there other countries that are building out data centers at scale and participating in that investment boom in AI.
- 8:02
- Unidentified speaker (UNKNOWN): Yeah.
- 8:02
- Unidentified speaker (speaker_2): it's a great question. Um, we're looking at these numbers using our real-time proxies of cross-border investment spending, and we're looking also at what national statistics tell us about where data center investment um happens domestically. Um, and um what's quite striking is that it's very broad based. It's really everywhere across the Asia Pacific region.
- 8:22
- Unidentified speaker (speaker_2): And if you're looking at the cross-border flows in isolation, it's it's increasingly the case that the spending that mostly originates in the US, uh, you know, the money coming from the big tech giants, is largely making its way to the Asia Pacific region. And that makes sense if you think about, you know, growth being here, right? Asia Pacific is still the fastest growing part of the world economy.
- 8:43
- Unidentified speaker (UNKNOWN): But
- 8:43
- Unidentified speaker (speaker_2): also Asia Pacific having some fairly large and high income consumer markets that um could potentially support um AI ventures. And I'll guess we'll talk about uh the business plans behind all of that and the issues that come with
- 8:56
- Unidentified speaker (UNKNOWN): that. But
- 8:57
- Unidentified speaker (speaker_2): coming back to your original question, Gourav, I think for Asia Pacific, the key factor, the key channel by which the AI boom is lifting economies is really by way of trade, right? Um investment is of course a big part of uh the AI build-outs, but um most countries don't make the chips. that are needed to fill data centers. So they
- 9:17
- Unidentified speaker (speaker_2): import chips from Korea, from Taiwan, from places in Southeast Asia. So it's the places that actually make these chips that fill the data centers that benefit. And looking at the latest numbers, they're truly staggering. John was reporting South Korean exports the other day, and I think uh exports year over year are somewhere north of 90%.
- 9:37
- Unidentified speaker (speaker_2): Um Singapore is north of uh north of 60, close to 70 if I recall correctly, and the same is true with Taiwan and that's after years of growing at a similar, similar pace. So um that just shows you how it's not really it's not really only a store of equities anymore, right? It's not the AI boom moving only financial markets. It's
- 9:57
- Unidentified speaker (speaker_2): moving it's moving entire economies. But the economies that it lifts the they're sort of like all
- 10:02
- Unidentified speaker (UNKNOWN): the they're very
- 10:02
- Unidentified speaker (speaker_2): very separate, right? There's many economies that benefit, but then there's also a huge number that don't really. So the splits between the two is quite remarkable as well.
- 10:11
- Unidentified speaker (UNKNOWN): You
- 10:11
- Unidentified speaker (speaker_0): know what I'm getting a sense of from here from listening to both of you? You don't really want to talk about the build-out of data centers. You want to talk about the exports and the financial market side of things. And I'm sensing that that's probably because you don't think that the first bit's that important.
- 10:24
- Unidentified speaker (UNKNOWN): I
- 10:25
- Unidentified speaker (speaker_0): just want to flush that out. Is that first bit not that important across the Asia Pacific region? Is it really the case that investment into building out data centers, that yes, it's happening, as you're saying, as you're saying here, Stefan, that some of it is coming through companies in the US investing in the Asia Pacific region? And as John says,
- 10:44
- Unidentified speaker (UNKNOWN): some
- 10:44
- Unidentified speaker (speaker_0): of these countries, countries like Australia, actually you have people using and adopting AI really rapidly and it's at scale, but still that's not what's moving the needle. Is that right?
- 10:55
- Unidentified speaker (speaker_2): Well, I'm not sure what you're thinking, John, but I would say at the moment that's how I would look at it, right? I mean the investment part is important. It's just that in terms of how it affects uh GDP, it's that the trade effect is really dominating at the moment. Again, to And just give you a few numbers here, Guraf, Guruf, we have Taiwan growing at double digit rates this year, Singapore, Malaysia growing north of five
- 11:15
- Unidentified speaker (speaker_2): percent, which is faster than China. So it's quite remarkable how much the trade numbers alone have shifted the picture. But you know, to sort of qualify that a little bit, of course the investment part of it is is important as well. It's just that the trade numbers are dominating right now, is the head of
- 11:30
- Unidentified speaker (UNKNOWN): the I I love it. I
- 11:30
- Unidentified speaker (speaker_0): love it. You just won't stop talking about the export side of things and the impact of on growth, no matter how hard I try to talk about. The build out of data centers is it's like, nah, let's move on, let's talk about the export side of things and the impact on growth. Okay, point taken. I'm gonna talk about the export side of things and the impact on growth. But you've told me all of that already, right? Um it's clearly, it's clearly very, very powerful.
- 11:50
- Unidentified speaker (speaker_0): If you were to rank order these economies in Asia Pacific, what which are the top three or four economies that have really um benefited from the export boom?
- 11:59
- Unidentified speaker (speaker_2): Well, uh I would put Taiwan at the top because the numbers in Taiwan are
- 12:03
- Unidentified speaker (UNKNOWN): really
- 12:03
- Unidentified speaker (speaker_2): through the roof. Yes, exactly.
- 12:05
- Unidentified speaker (UNKNOWN): Um
- 12:07
- Unidentified speaker (speaker_2): actually the next three are all close to one another: Singapore, Malaysia, uh, Korea. Um, but the sort of the timing of the boom varies a little bit. But those are uh I would say the four economies that are at the top if you look purely at the trade numbers.
- 12:22
- Unidentified speaker (UNKNOWN): Okay,
- 12:22
- Unidentified speaker (speaker_0): so those are the top four economies. Now people also talk about China
- 12:26
- Unidentified speaker (UNKNOWN): and
- 12:26
- Unidentified speaker (speaker_0): it's one trillion surplus over the course of ten months. And and the general feeling is that that's been helped a lot by China's own exports of chips. And they also talk about Japan. But to what extent uh uh uh are these countries participating in in the export boom?
- 12:42
- Unidentified speaker (speaker_2): Well, China is a place that makes a lot of electronics. Uh but with China, of course, you have that uh the caveat that you know the most advanced chips you can't really you can't really export to China. So it's a key part of the overall value chain. It also produces lower-tier chips that feed into electronics more broadly, but it's not sort of
- 13:00
- Unidentified speaker (UNKNOWN): at
- 13:00
- Unidentified speaker (speaker_2): the vanguard of the whole thing, right? It's not sort of producing the bleeding edge sort of chips that go into the whole, into the whole boom. Japan is structurally different in that its companies operate abroad, right? Uh Japanese firms produce all over the world, but it doesn't really export all that much.
- 13:15
- Unidentified speaker (speaker_2): from its own shores. So it doesn't benefit from the EI boom the way regional neighbors do. And the trade numbers are looking quite a bit more modest for that reason.
- 13:24
- Unidentified speaker (speaker_0): Yeah, but with China, I suppose it's also such so much more of a diversified exporter and such a big exporter. It's probably hard to have one sector alone move the needle that much. Um But I'm I'm not so sure about what you said about it not making
- 13:38
- Unidentified speaker (UNKNOWN): you know the
- 13:38
- Unidentified speaker (speaker_0): leap. bleeding as chips, which it doesn't. It's not that bit that I disagree with. It's it's also that a country like Malaysia doesn't, and yet it moves the needle in Malaysia. So clearly that is a country where I guess it's it's really focused on that sector. And then we've seen a massive boom in that sector, even if it's not bleeding as chips, because we know that data centers require all sorts of equipment
- 13:58
- Unidentified speaker (speaker_0): and and all sorts of chips. Um but in the case of China, it's just not that noticeable, I guess. That's what you're saying.
- 14:05
- Unidentified speaker (speaker_2): Absolutely. I think uh we've may have given this example on a previous podcast, but there's a single state in Penang which makes most of the country's electronics
- 14:13
- Unidentified speaker (UNKNOWN): and
- 14:14
- Unidentified speaker (speaker_2): those now account that the exports from that state now account for over half of the country's overall exports. So that goes to the point you make, Groof, about Chinese exports These just being a whole lot more diversified, right? So in part it's it's that, right? It depends on how you look at it in terms of the absolute numbers. China still matters, but it's just that it it dominates overall when it comes to exporting.
- 14:32
- Unidentified speaker (speaker_0): Yes, exactly. And that reminds me of a great paper that you put together not so long ago called Made in China, talking about how it's how China has managed to diversify its exports so enormously over a period of the last 20 years or so, and now it is pretty much the world's manufacturer. But that's not what we're here to talk about today. It's really great that these economies have are benefiting from
- 14:52
- Unidentified speaker (speaker_0): the boom in AI and build out of data centers by exporting into all of this. But what about the rest of their economies? It's it feels like what you would want at this point is for if a sector is
- 15:06
- Unidentified speaker (speaker_0): dragging the economy along, it would be quite nice if the performance of that sector spilled over to other parts of the economy and it actually led to consumers feeling flush with cash jobs. being generated in other sectors, the non-traded sectors, if you like, also then benefiting. Is that happening? Or is this really very narrow and it's just the this this one sector that's driving growth and
- 15:25
- Unidentified speaker (speaker_0): it's the economies are getting kind of skewed?
- 15:29
- John Bromhead: Well, what I would say, I mean, let's take you know, Korea, which is one of the biggest beneficiaries of the AI stuff. So probably a good case study. Um
- 15:40
- Unidentified speaker (UNKNOWN): and
- 15:40
- John Bromhead: I think this is a massive question because you got we we we got this recent Q3 GDP print and we saw private consumption essentially flatline. So that hasn't happened in nearly four years. So
- 15:52
- Unidentified speaker (UNKNOWN): you
- 15:52
- John Bromhead: know, th this is the I guess the right question to ask is like how how does the AI kind of story actually flow through the households? And there's two kind of main channels, I guess. One is higher corporate profits should flow through to the uh government's tax revenue. Uh that should support government spending. And then there's the kind of wealth effect which
- 16:12
- John Bromhead: we spoke about a little bit before. Uh Samsung and SK Heinex have delivered enormous returns for their sh shareholders. Um
- 16:21
- Unidentified speaker (UNKNOWN): I
- 16:21
- John Bromhead: guess one issue is that uh some of that wealth effect does seem to be concentrated among wealthier households and institutions. So they're kind of less likely to spend that extra wealth.
- 16:32
- Unidentified speaker (UNKNOWN): But
- 16:32
- John Bromhead: the other, I mean, the other fact is, I mean, and this is the same everywhere, the Bank of Korea is hiking rates, higher fuel prices, that all hurts. And so, you know, I think for a lot of households, and this is maybe not just for Korea, um, really feeling the pain and not really feeling the benefit.
- 16:51
- Unidentified speaker (UNKNOWN): And
- 16:51
- Unidentified speaker (speaker_0): I suppose these sec this the sector does not also uh employ a lot of people. It's just not that labor intensive. So you're not going to see an effect from people earning lots of money and going out and spending that money and and helping drive up services, for instance.
- 17:06
- Unidentified speaker (UNKNOWN): I
- 17:06
- Unidentified speaker (speaker_2): love that you mentioned labor markets, Groov, because I was just thinking about Korea in that context. There were these reports earlier this year. and late last year, I believe, of workers uh in areas that are industries that are plugged into the AI boom receiving record bonuses. And uh over the last couple of weeks have seen
- 17:26
- Unidentified speaker (speaker_2): more media reports pointing out how that's creating some tension, some friction in that you know, um within the country, it's not great for one share of workers, a very small share, a very small fraction of workers to be given these sky high bonuses when the rest of the economy isn't doing isn't doing so well because everything that is not plugged in the AI
- 17:46
- Unidentified speaker (speaker_2): boom right now, if anything is sort of suffering from the negative fallouts, contributes to inflation, it absorbs all of the available construction capacity, et cetera, et cetera, are things that we've talked about previously. So there is that aspect to it as well where it creates that disparity within countries, right? It's winners and losers. There's the stark gap between them as
- 18:05
- Unidentified speaker (speaker_2): well.
- 18:06
- Unidentified speaker (speaker_0): Sense of winners and losers. That's actually why I raised this point because I'm I've been following this and I've heard all these stories of a small fraction of the labor force that's employed in these sectors actually getting these disproportionate bonuses and people not being very happy more generally about it. Um
- 18:23
- Unidentified speaker (speaker_0): so overall, I think the point John was making was that. and both you are both of you are making, is that the economic benefits are really staying within this sector. So it's additive. There's not much in the way of synergistic gains across the economy. And that's not very good for other sectors of the economy, which are just languishing.
- 18:43
- Unidentified speaker (speaker_0): Households are not feeling feeling the gain from all of this.
- 18:47
- Unidentified speaker (speaker_0): Wealthier households are investing in stocks, but they are not. they're not the people that are likely to spend much anyway. So that doesn't help. It's not a very labor-intensive sector. So you don't see big labor gains. You don't see spillover benefits to other sectors. That's a little bit troubling. And I think that statistic you came up with, John, that if in third-quarter GDP growth,
- 19:07
- Unidentified speaker (speaker_0): if you stripped out all the AI-related AI related components, the number would go down from 1% quarter and on quarter to 0.2% quarter and quarter. on quarter. So, in a way, if you think of that as the rest of the economy, that's not a great performance, right?
- 19:23
- Unidentified speaker (speaker_0): But still, these these economies are doing very well. And over the course of this year, we've actually raised our growth forecast for many of these economies just off the back of this. So it's a remarkable, it's a truly remarkable performance. Um
- 19:36
- Unidentified speaker (speaker_0): You talked about
- 19:37
- Unidentified speaker (UNKNOWN): some
- 19:38
- Unidentified speaker (speaker_0): people getting wealthy.
- 19:39
- Unidentified speaker (speaker_0): in these countries. Now they're getting wealthy because the stock market's been doing so well, and some of these Asian stock markets have just skyrocketed this year. Some of it's come off in recent months. We know that. The COSPI, for instance, our favorite stock index to monitor, that's um that that reached stratospheric Atmospheric levels and and sort of come back quite
- 19:59
- Unidentified speaker (speaker_0): far down from the high heavens in the last few months. Um
- 20:04
- Unidentified speaker (UNKNOWN): what
- 20:04
- Unidentified speaker (speaker_0): do you think is the wealth effect?
- 20:07
- John Bromhead: I I would say that like maybe looking at equity markets is probably one of the clearer signs that the AI trade has started to change, um, particularly in Asian equity markets. It's just since the start of October, and I raised this because you mentioned the Cosby, but Cosby's down over 20%. Uh Nikkei's down eleven 11%. percent, and it was was it
- 20:27
- John Bromhead: was worse a few days ago. Taiwan's down eight to nine percent. Uh China AI kind of hardware has been hammered. Um the Philly semiconductor index is down over 20%, or it was. So,
- 20:43
- Unidentified speaker (UNKNOWN): and
- 20:44
- John Bromhead: that's just the public markets. I mean, if you look at the private markets, um, and this is maybe something we can talk about as well, but Anthropic taking a massive haircut on the valuation that it was looking for as we head into IPO next week. They were looking for 2 trillion. That's now going to be 1.3 or
- 21:03
- John Bromhead: 1.4 trillion, still enormous, but a 30 to 40% decline in just a couple of months.
- 21:11
- John Bromhead: So what is interesting is you look at somewhere like India, its stock market's up kind of 12% in the same time period. So
- 21:21
- Unidentified speaker (UNKNOWN): there
- 21:21
- John Bromhead: is a kind of interesting divergence happening here where, and I guess it's a little bit similar to what we saw in the US with like Apple outperforming the other hyperscalers, the market's starting to look for the places that aren't i as exposed to the AI capex cycle. And I think we're increasingly seeing this rotation. away from the regions
- 21:41
- John Bromhead: and the companies kind of selling the AI dream and more towards those delivering growth and earnings um maybe without the AI baggage.
- 21:49
- Unidentified speaker (UNKNOWN): Yeah,
- 21:50
- Unidentified speaker (speaker_0): so some of the shines coming off AI. Now that's not causing um serious concerns about a financial bust yet. But certainly it's it's it's interesting to see this correction happening. And it's really um been about, I guess, people thinking about the broader consequences of this AI build out from um
- 22:10
- Unidentified speaker (speaker_0): the applicability of AI, the productivity gains from AI, to um the resource requirements and and and social pushback, something you talked about at the start of this show, John.
- 22:21
- Unidentified speaker (UNKNOWN): Um
- 22:22
- Unidentified speaker (UNKNOWN): now
- 22:23
- Unidentified speaker (speaker_0): this is this is also a global correction. So these Asian stocks are falling also because they're their US counter counterparts are falling, anthropics falling, it's et cetera, et cetera. And and the latest episode with uh NVIDIA doesn't help, does it? Now that's also taken a fair bit of shine off NVIDIA stocks. That may just be something, um, an isolated thing and it's it's it's
- 22:43
- Unidentified speaker (speaker_0): it's it's it's limited to one company and it passes over time, but it doesn't help.
- 22:47
- Unidentified speaker (UNKNOWN): So
- 22:48
- Unidentified speaker (speaker_0): this is really a a global correction in AI stocks and and and these Asian markets are also taking a beating. It but it's interesting the point you make that people are perhaps finding finding havens elsewhere in a market like India is doing quite well as a result.
- 23:03
- Unidentified speaker (UNKNOWN): Yeah,
- 23:03
- John Bromhead: and I th I think like, you know, increasingly um people are becoming comfortable that
- 23:11
- Unidentified speaker (UNKNOWN): you know
- 23:11
- John Bromhead: adoption is climbing. I think the initial validation tests are uh have have clearly been passed. So it's now very much more about um who's going to finance this, the cost of capital, uh, the politics of power and land,
- 23:26
- Unidentified speaker (UNKNOWN): um,
- 23:26
- John Bromhead: and whether or not um the market actually believes in the returns. Um you know, and I think what's interesting too, and and and we mentioned it at this at the top, but like one of the big catalysts for this. recent move is um the midterms in the US. And data centers and
- 23:45
- John Bromhead: development were on 48 specific ballot measures across 10 different states, ranging from kind of outright bans to all the way through to kind of just slowing development. And 47 out of 48 lost, or effectively, you know, um people voted very handily against more data
- 24:05
- John Bromhead: centre investment. And so what's interesting is that Democrats have now won governorships in some of the states at the very kind of heart of the US data centre boom. Um Texas alone accounts for I think twenty-five 25%, percent, like roughly a quarter of of US data centre center capacity through
- 24:25
- John Bromhead: to twenty 2028. twenty eight. So
- 24:27
- Unidentified speaker (UNKNOWN): if
- 24:27
- John Bromhead: you if you if you add on Ohio and Georgia on top of that, it's probably closer to 50%. And so you've got these democratic governors coming in, you know, essentially campaigning on winding back or slowing down investment. And so I think just generally that pipeline, and and and I guess why we
- 24:47
- John Bromhead: talked about kind of data center investment in Asia, not worrying, not making mattering as much, is because the sheer value of CapEx in the US is so far above kind of everywhere else in the rest of the world that it it's it's not as significant. Um but the fact that the pipeline there, the place where it's all happening, is
- 25:07
- John Bromhead: facing
- 25:08
- Unidentified speaker (UNKNOWN): kind
- 25:08
- John Bromhead: of more uncertain political and and regulatory environments, um, I think you know that that that is going to matter.
- 25:14
- Unidentified speaker (UNKNOWN): And
- 25:14
- Unidentified speaker (speaker_0): that is going to matter to also these companies in Asian economies that are selling into this into this boom, because we know that these companies are very solid order books, don't we? And do we are we now concerned about their order books? Well, we are concerned about the order books. I guess the question is how concerned should we be about their order books?
- 25:33
- Unidentified speaker (UNKNOWN): Well
- 25:33
- John Bromhead: not no it's so so I think that the main thing is pro it's not going gonna to be a twenty twenty seven issue. It's going to gonna be a kind of twenty twenty eight twenty twenty nine, twenty thirty thing. Um so I I yeah, I mean to the extent that um the chip chip manufacturers are s uh are sold out for at least the next eighteen months, it's that's not gonna be an issue. Uh
- 25:53
- John Bromhead: but to the extent that that kind of future pipeline looks a little bit more shaky, I think like that that's certainly worth looking at. Um I I mean there's a debate. I mean Stefan and I were having this conversation about whether or not a slowdown in US data center investment is necessarily a bad thing. I mean,
- 26:13
- John Bromhead: there's
- 26:15
- John Bromhead: because you know, I I think there's already some concerns about the kind of unit economics
- 26:21
- Unidentified speaker (UNKNOWN): of
- 26:21
- John Bromhead: running these data centers, and to the extent that you have a slightly like slower possible. pipeline, it actually might improve um the economics for the existing providers. So you might see companies that already have spare computing capacity, you might see their value rise. Um we've already seen a lot of concerns
- 26:41
- John Bromhead: around the neo clouds and the,
- 26:43
- Unidentified speaker (UNKNOWN): I
- 26:43
- John Bromhead: suppose, more leveraged providers of these data centers. Um and it, you know, a and at the margin, it also might have you know help cool inflation a little bit. Um, to the extent that you're not pouring all these electricity labor materials into into building. Um, perhaps that helps the Fed not have to hike a bunch of times. So
- 27:03
- Unidentified speaker (UNKNOWN): a
- 27:03
- John Bromhead: debate maybe worth having.
- 27:06
- Unidentified speaker (UNKNOWN): And
- 27:06
- Unidentified speaker (speaker_2): Love Hawk, we keep coming back to data center KPEx. Uh perhaps the earlier reply we gave to your question, Gurev, was a bit too wishy-washy in in terms of saying that, hey, you know, it matters a little bit when maybe we should have we should have been more clear and say, no, it doesn't actually all matter all that all that much. But just to stay on that topic a little bit, um, because it's something,
- 27:26
- Unidentified speaker (speaker_2): of course, we are very concerned about in Asia Pacific, right? What would happen if the AI boom were to turn into busts, even just a moderate correction, might see output across a number of economies falling. Taiwan GDP
- 27:39
- John Bromhead: fell for a couple of quarters coming out of the pandemic when everybody was done buying tablet computers for the kids because they suddenly had to stay home to study, et cetera, right? So it wouldn't it wouldn't take much to see repeat of that sort of scenario, especially given how elevated all of the numbers are right now. But to elaborate on the point that John made there earlier,
- 27:59
- John Bromhead: right? If you get slowing data center K KPEX bags in the US part of the KPEx might just move elsewhere, right? It might move to Asia Pacific, it might move to other countries because the geopolitical pressures behind all of that are still there, right? There is competition for leadership in that space. And AI matters enormously from
- 28:19
- John Bromhead: a strategic point of view for countries' politics, for security, etc., et cetera, which is an underlying driving factor that will still be there tomorrow. regardless of whether we're able to work out the economics or not. And if I might just sort of draw a parallel here with the two thousands dot com bubble, there were a whole bunch of econom a whole bunch
- 28:39
- John Bromhead: of co uh uh companies that benefited from the dot com boom in the late nineteen nineties that drove all of the investment in communications infrastructure, et cetera, that are not around anymore today. but the infrastructure is still there and the communications build-out continued. So to the extent that it's Asia that makes the hardware that goes into all of this, I guess I'm less
- 28:59
- John Bromhead: concerned about Asia because of that, right? It's just the rest of the world hasn't learned to make chips at the same level of sophistication that Taiwan and Korea have yet. So AI is going to be around tomorrow and it's still Asia making the chips. The transition period from this very sort of from the period where you've got these very elevated
- 29:19
- John Bromhead: asset price valuations to a period where you go into more sustainable long term growth is going to be painful. But I guess I'm not sort of worried about Asia from a real economy point of view. It's more about the financial market aspects of it that I'm worried about, right? Because that could cause a lot of pain along the
- 29:32
- Unidentified speaker (UNKNOWN): way.
- 29:33
- Unidentified speaker (speaker_0): Those are great points that Asia makes chips. And actually nobody else can make chips the way Asia makes. chips. So that just means that this is going to continue for Asia.
- 29:44
- Unidentified speaker (speaker_0): It's an interesting point about whether data center investment could just go to other countries, relocate to other countries if there's pushback in the US. I think it could happen, but there's also a bit of a question mark around that from if if you think about what John was saying at the outset here, that a country like Australia, where there's been a fair amount of interest in building out data centers. now there's pushback there as well. So
- 30:04
- Unidentified speaker (speaker_0): I think that's an open question. Also in other countries, if you think about their grid capabilities, their other their water capabilities, it might you might rub rub up against some pretty significant barriers in pushing data center investment out to other countries. But I'm I think you're right in that there will certainly be a desire to do
- 30:24
- Unidentified speaker (UNKNOWN): so.
- 30:25
- Unidentified speaker (speaker_0): And then that might that might help. um spread the data center investment both across to other countries and also keep it going for a for a while. Let's come back to that financial and markets aspect of it because that worries me. So far it's been fine. We've seen a fair bit come off stocks um in in the AI related stocks, but that hasn't
- 30:45
- Unidentified speaker (speaker_0): translated into broader financial pain to other markets, to other sectors, to other asset classes. But I still remain worried that this could happen. And if for no other reason that financing conditions are generally tightening, we talked a bit about or briefly mentioned inflation tightening monetary policy. All of this is leading to
- 31:05
- Unidentified speaker (speaker_0): higher debt costs, there's a lot of uh lot of debt that's gone into financing data centers. Um generally funding conditions are tightening, risk appetites tightening. So it doesn't take much of a blow to see the current correction in AI stocks become a bigger correction, spill across to other asset classes, and generally create some kind of financial market
- 31:25
- Unidentified speaker (speaker_0): dislocation, which I feel would be very globally painful.
- 31:28
- Unidentified speaker (speaker_3): Yeah. Yeah. Look, uh I I I think I maybe add two points on this, right? Um
- 31:34
- Unidentified speaker (speaker_3): part of what has made this such a virtuous cycle on the way up is that we have these kind of self reinforcing dynamics
- 31:45
- Unidentified speaker (speaker_3): inside of the financing of all this stuff. So a lot of the deals and like, you know, companies that are buying chips to to to put into these data centers, a lot of that's funded by debt. And increasingly it's backed by the very E AI ecosystem that it's financing. So a Google or an Anthropic
- 32:05
- Unidentified speaker (speaker_3): or any of these kind of data center builders sign a multi billion dollar contract, compute contract.
- 32:13
- Unidentified speaker (speaker_3): Sorry, I should say the the users of the data centers will sign a contract with a data center provider like a Neo Cloud. That Neo NeoCloud Cloud then goes out and uses that contract to secure. more financing. And so it then buys the Nvidia chips and builds the data centers. So
- 32:33
- Unidentified speaker (speaker_3): basically, one company's spending commitment becomes another company's collateral. And that unlocks more lending and more spending. And you can kind of see how you get this virtuous cycle on the way up. Very powerful. On the way down, though, the kind of process can reverse. And so if demand disappoints, you get lend. lenders question future cash flows,
- 32:53
- Unidentified speaker (speaker_3): collateral values, borrowing costs rise. And you have already seen that. I mean, CVS kind of blowing out on a lot of these neo-cloud providers, Core Weave in particular. But
- 33:06
- Unidentified speaker (UNKNOWN): the
- 33:06
- Unidentified speaker (speaker_3): other thing I would say, just on I guess on the financing side, but like one of the other reasons that this AI kind of story has been
- 33:14
- Unidentified speaker (UNKNOWN): I
- 33:15
- Unidentified speaker (speaker_3): suppose, so resilient is that it's been supported by extraordinary profit numbers. And and particularly in the US, what makes it interesting is that like a decent chunk of those earnings uh themselves are a function of rising AI valuations. So
- 33:32
- Unidentified speaker (UNKNOWN): if
- 33:32
- Unidentified speaker (speaker_3): you're you Well, know, a bunch of the big tech companies basically have huge positions in the frontier labs, like Anthropic and OpenAI. And what happens over time is that those valuations go up and Google and Amazon and whatever get to revalue their positions and basically report it as profits. And
- 33:52
- Unidentified speaker (speaker_3): so we we've seen this across the industry in various different ways. But um what's going to be interesting, I guess, is when you have private companies like Anthropic now being become public, which they will be next week, suddenly these kind of uh mark-to-market uh valuations happen every day, every second, right?
- 34:12
- Unidentified speaker (speaker_3): Like you're publicly listed now. Uh and so um
- 34:17
- Unidentified speaker (UNKNOWN): I
- 34:17
- Unidentified speaker (speaker_3): think that could now be a source of volatility rather than this like a creative earnings thing that we've seen on the way up.
- 34:24
- Unidentified speaker (speaker_0): And also I suppose what you're seeing saying here is that what's what starts out as a virtuous cycle can end up as a bubble, right? And that's the problem here. And a bubble getting pricked can be very, very nasty. But so far, what has been happening hasn't affected financial markets more broadly. So there is hope here that we get in a baseline view of the world, some of that
- 34:44
- Unidentified speaker (speaker_0): froth comes off. And as you as you were both saying, it gives this industry sector a bit more time to reflect on what is sustainable. what is good for long-run growth, and better, better, more robust financing structures coming along. And that actually continues to power growth without any dislocation. The bad scenario is one where that dislocation occurs. And it does feel as though the pain would
- 35:04
- Unidentified speaker (speaker_0): be intense at this point because of all the other problems that the global economy is facing. But don't forget, we've still got higher tariffs. We still we don't talk about this stuff anymore, but there's still higher tariffs. right? Then there's still we have high energy prices, we have rising inflation, we have a global we have global monetary tightening, we have a lot of debt
- 35:22
- Unidentified speaker (UNKNOWN): um
- 35:23
- Unidentified speaker (speaker_0): in the system. All of this could be quite nasty, and this could indeed be the trigger for all of that if it gets gets worse. But maybe end end on a longer note here and something that's that's more positive. If you
- 35:36
- Unidentified speaker (UNKNOWN): if I
- 35:36
- Unidentified speaker (speaker_0): bring it back to this region, and this is a really broad based question here, and think about the long term for AI. Now you talked about this, Stefan, saying that in the long term AI still may um Asia still makes chips. I beg your pardon. Asia still makes chips and therefore you're not concerned about Asian growth. But what about the productivity and adoption side of things? What does it take for
- 35:56
- Unidentified speaker (speaker_0): AI to be embedded into Asia Pacific economies and to really drive stronger productivity growth in in in the longer term. To what extent do you see that happening?
- 36:06
- John Bromhead: That's a really good question as well. I don't know that that aspect of the boom is any more clear in Asia Pacific uh than it is elsewhere, right? We're getting a lot of anecdotal evidence about how it's affecting labor markets in a way that is not exactly conducive to social stability. Young people struggling to find jobs and uh, you know, firms trying
- 36:26
- John Bromhead: to to replace uh knowledge work sort of still at the lower end of the scale by way of AI agents, etc. So that might be productivity enhancing, that might be good for the bottom line. But to our discussion earlier, right, it's um it's not great from a long term point of view if you think more about the social equity aspects of it because it's just one
- 36:46
- John Bromhead: part of the economy that benefits and the rest's rest the rest doesn't so much. So it's not very clear yet. It's just that it you know we're starting to see early signs that this is an aspect by which it it is lifting productivity but not exactly helping economies at large, if that makes sense.
- 37:01
- Unidentified speaker (UNKNOWN): And
- 37:01
- Unidentified speaker (speaker_0): then one of the things that worries me about the build out of of AI in into say the middle income economies of Asia, not so much the high high-income income economies, but much more around the middle income and lower income economies of Asia is that these are also not economies that have the capacity to think about reforming, say, the tax systems or their welfare
- 37:21
- Unidentified speaker (speaker_0): systems that much to be able to help people that lose jobs. So we talk about uh a minimum minimum wage, and we talk about universal basic income, and we talk about Finland experimenting with things with schemes like this. We don't talk about Asian economies experimenting with schemes like this. So actually you could see fairly nasty effects from a labor market point of view, which of course leads to
- 37:41
- Unidentified speaker (speaker_0): huge big social issues.
- 37:43
- John Bromhead: Yeah, absolutely. And that's I think a really good point in that the benefits, but also the sort of negative implications of the AI rollout, they apply globally, right? They don't necessarily stay in the economies that run the AI labs uh where the frontier labs are located, because everybody has access to those, right? So whatever problem.
- 38:02
- John Bromhead: productivity benefits you're going to see, they're going to show up everywhere. And to your point, I'm quite concerned about what it does to the economic model in South Asia or places like the Philippines that in the past have relied on services exports, IT exports, to grow in a world where the sort of old export-led goods
- 38:22
- John Bromhead: export-led growth model is not quite as accessible as it used to be, right? I think those economies. are certainly at risk. And those are the ones that I would be worried about, medium to longer term.
- 38:31
- Unidentified speaker (speaker_0): Yes, and and again, it's unclear. I think I I need to really really stress that point that anyone trying to make a very strong story out of any of these issues will find themselves rubbing up against big doses of uncertainty. Um and it's it's something we're just gonna have to wait and see how this works out, because even with the those service sectors, it could be a pretty
- 38:51
- Unidentified speaker (speaker_0): long term story before those jobs go in those service sectors. You'd think that they'd be vulnerable. Um, India is a good example, right? It's you'd think that with its business process outsourcing sector, which is pretty, pretty big, it would be immediately affected by AI adoption. But so far it hasn't. Now it could happen. We I I remain just as concerned as you are about
- 39:11
- Unidentified speaker (speaker_0): countries in Asia that have um service sectors. and that they export and these are export sectors. But so far it's not happening, but it's a big risk for the medium term. And the extent to which it is a risk and the extent to which this risk crystallizes really remains to be seen. So lots of uncertainty there. Probably probably a final note to end on is one that I've
- 39:31
- Unidentified speaker (speaker_0): been thinking about as uh we've been talking about this. It feels like
- 39:35
- Unidentified speaker (UNKNOWN): there's
- 39:35
- Unidentified speaker (speaker_0): a bit of asymmetry here on the way up. Um
- 39:39
- Unidentified speaker (UNKNOWN): the
- 39:39
- Unidentified speaker (speaker_0): AI boom has really benefited certain Asian economies through the exports, through exports of chips and equipment and involved in the build-out of data centers. It hasn't really spilled over to other sectors of the economy. So it's a pretty much a sectoral story. On the way down, we're concerned here about financial market corrections. Over the medium term, we might be concerned about job
- 39:59
- Unidentified speaker (speaker_0): losses. These could actually spill over very rapidly. into other sectors of the economy and cause severe dislocation. So this asymmetry worries me. On the way up, it's not been that great in benefiting other sectors of the economy. If there is a rapid correction or some negative effects from job losses, that could affect other sectors of the economy, and that could be
- 40:19
- Unidentified speaker (speaker_0): pretty bad.
- 40:21
- Unidentified speaker (UNKNOWN): So
- 40:22
- Unidentified speaker (speaker_3): like yeah, I think the context for which all of this AI stuff is happening matters a lot. And obviously, like economies are coming into this with a very different starting point. Um the UN
- 40:33
- Unidentified speaker (UNKNOWN): uh
- 40:34
- Unidentified speaker (speaker_3): development programme put out a a a piece of research yesterday talking about their projection for one point two billion young people in emerging and developing economies coming into the workforce for the next 10 to 15 years,
- 40:48
- Unidentified speaker (UNKNOWN): uh,
- 40:48
- Unidentified speaker (speaker_3): but only projecting around 400 million jobs. uh for those people. So, you know, a very big shortfall already into an environment that um, as Stefan mentioned, knowledge work looks a little bit shaky. Um,
- 41:02
- Unidentified speaker (UNKNOWN): you
- 41:03
- Unidentified speaker (speaker_3): know, you have to say that's not a particularly optimistic um outlook. I I'd prefer to maybe finish on something a bit rosier.
- 41:12
- John Bromhead: Let me maybe uh take
- 41:13
- Unidentified speaker (UNKNOWN): the opposite
- 41:14
- Unidentified speaker (speaker_2): point of view.
- 41:15
- John Bromhead: because I think it's important to remember that what we're discussing here is still largely hypothetical, and to that extent, we talked about risks, right? Our baseline view for the region is still that the AI boom is going to moderate, and for that reason, growth next year is also going to moderate across exports, investment spending, GDP, etc. Um, one
- 41:35
- John Bromhead: reason is that the AI boom is creating a number of distortions, but more fundamental. fundamentally, I think the view is centered around the fact that right now it's just not clear what the business models of these AI ventures are going to be, right? The frontier labs have no modes, as it were, right? It's not clear that you can't just easily switch away from them tomorrow to open open
- 41:55
- John Bromhead: weight models. It's not clear that the depreciation schedules are going gonna to work out. So, long story short, um, we're not sure that. the models are priced correctly or that we're at equilibrium yet, because right now you're just getting AI at sort of bargain-bin prices. Uh to our in our
- 42:12
- John Bromhead: in our thinking, essentially, there needs to be some sort of correction. Uh and I think that correction could be good if it helps price the whole thing a little bit more fairly, and then maybe the implications wouldn't won't be quite as far reaching, right? Because we're sort of thinking about a world where AI is essentially free, but
- 42:27
- Unidentified speaker (UNKNOWN): I
- 42:27
- John Bromhead: don't think that is the that is going to be the case tomorrow, right? Um the pricing will need to change and once the pricing changes, it will become clear that certain things, uh that there are certain things that AI isn't ready yet for, and that maybe the social implications won't be quite as dire as we've as we've discussed.
- 42:45
- Unidentified speaker (speaker_0): Also I I'm going to take the long view of technology here, right? Technology's always been beneficial. for human development and economic growth. Well, most technologies, not every technology, right? Um, but most technologies are beneficial for human development and and economic growth. And there's
- 43:01
- Unidentified speaker (UNKNOWN): you know
- 43:02
- Unidentified speaker (speaker_0): there are there are reasons to believe that AI will be the same. Um we don't know how the economy will shift, what sort of new jobs might come out. We don't know um what sort of trade patterns will emerge. There's just so much uncertainty around so many big questions. It's too early to conclude that AI is negative. Um it's too early to include to conclude that AI
- 43:22
- Unidentified speaker (speaker_0): is is um hugely positive. The jury's out.
- 43:27
- Unidentified speaker (UNKNOWN): And
- 43:27
- Unidentified speaker (speaker_0): probably the bigger one of the bigger areas of concerns will be the usage of AI and what what what areas in which AI gets used and how that affects things like information flows.
- 43:38
- Unidentified speaker (UNKNOWN): Um
- 43:38
- Unidentified speaker (speaker_0): how AI gets regulated as a result. We just don't know. So I'm gonna leave it there. Lots of things we don't know. Um we talked a lot about the short term and and and how AI is benefiting economies in the Asia Pacific, and we've got concerns around valuations and what that does to financial wealth and what that might mean if if it
- 43:58
- Unidentified speaker (speaker_0): there's a big financial market correction. So far I think everybody agrees on the this show that the correction that we've seen to date, these are not having big wealth effects, negative wealth effects, and big economic repercussions. So that's good news. And that's probably the best good news to end on. So thank you, Stefan. Thank you, John, for another great discussion. And of course, thank you to our listeners.
- 44:18
- Unidentified speaker (speaker_0): You've been listening to the Global Economy Unwrapped.