R.P. Eddy: Don't Ignore The Cassandras

R.P. Eddy · Ergo

AI job losses have already begun; Citrini's labor-collapse scenario increasingly plausible.

Eddy lists job loss among AI problems that have already arrived: 'Will we lose jobs? We've already begun to have that problem.' He cites the bank-of-experts style warnings that people will lose jobs as accountants, lawyers, dermatologists and Wall Street analysts, and says the claim that new jobs will replace them is unconvincing. He points to the Citrini report, which critics dismissed as self-promotion when it came out, and says it increasingly makes sense as a scenario in which systemic economic dependence on labor collapses. He ties this to the economy's growing dependence on the AI build-out.

“Will we lose jobs? We've already begun to have that problem.”at 16:27

R.P. Eddy · Ergo

AI will make cognitive excellence and knowledge less valuable as a differentiator for profit.

Eddy describes his children and their peers, who want to learn and can learn in ways he has never seen. He realises the cognitive excellence being invested in and trained will become less valuable once a machine can do it all, 24/7. That is a scary problem. He argues knowledge should be valued for its own sake and not as a tool for profit, because knowledge will not be a differentiator for profit in the AI era.

“that cognitive excellence that we're investing in and building and training, will be less valuable when there's a machine that can do all that 20 fourseven.”at 56:28

R.P. Eddy · Ergo

Expects US to restrict Chinese AI models in defense and finance; not bullish on them.

Eddy says he is not very bullish on Chinese models and expects US government restrictions on them. He says government will be explicit about not encouraging their use, if not banning them outright, and that defense contractors and financial markets will not be allowed to use Chinese models. He cites the Huawei playbook, where US restrictions shut a Chinese vendor out of sensitive sectors, as the template. He expects American frontier and cheaper models, plus future American open-weight models, to fill the gap.

“the U. S. Government is going to be highly explicit that at least not encouraging the use of Chinese models, if not outright banning the use of Chinese models. In fact, you should bank on that”at 24:35

R.P. Eddy · Ergo

The US should tighten China's AI access and chip exports rather than negotiate with China from weakness.

Eddy argues that if the US has so much trouble with China, it should hammer it harder. He asks why tariffs are not higher and why the US is not doing more to restrict China's capacity to build AI. He questions why the US lets China get ahead in AI at all, why China has access to data centres in Asia, and why chips, including high-power ones, are allowed to go there. He calls AI central to the broader contest.

“Why are we letting them get ahead in a I at all? Why are we Why you letting them have access to data centers in Asia? Why are we letting any chips go in there?”at 39:18

R.P. Eddy · Ergo

Iran has lost control of the Strait; US blockade and Operation Stranglehold are squeezing Iran.

Eddy argues that the headlines suggesting deadlock in the Strait are wrong. In mid-September Iran lost control of the Strait, and US oil flows through it are roughly the same as before the war. The US is also blockading Iranian exports, and no Iranian ship carrying oil has transited since July 26. That hits China, the biggest buyer of Iranian oil, though a large cargo about ten days earlier likely went to China. He says Operation Stranglehold, led by Scott Bessent, is working and Iran is under real pressure. He warns that Iran's leadership may still produce a wildcard because it feels it needs an extortive tool.

“In mid September, Iran firmly lost control of the Strait. The U. S. Started getting through oil at the same level, almost the same level we had before the war, and and were blockading Iranian export.”at 31:56

R.P. Eddy · Ergo

US–Iran nuclear deal is likely this year, with odds rising after the midterms.

Eddy says a deal before the midterms is unlikely. Trump wants a nuclear deal rather than another MOU, and the complexity of nuclear negotiations plus continued Iranian obstinacy means any acceptable deal is more than a month away, probably after the midterms. Eddy's team thinks a deal will come and will be nuclear, with probability rising toward year-end and after the midterms. He puts it at probably 60%-plus this year, rising as the new year approaches. He cites real diplomatic progress behind the scenes, Iran's economic pressure and Iranian leaders' awareness that Trump is unlikely to win the midterms.

“And I think we'd say, you know, it's probably 60% plus likely we get a deal in this year and starts to curb up and up as we get over the new year”at 36:04

Porter Collins · Seawolf Capital

US deficits near 6% of GDP are unsustainable; interest plus entitlements already exceed tax receipts.

Collins argues the US cannot keep running fiscal deficits of around 6% indefinitely, and that the reckoning will come at some point. He cites the figure that interest expense plus entitlements is already over 100% of tax receipts. Deficits at this level are not tenable, so the adjustment is inevitable even if its timing is uncertain.

“We can't keep running 6% fiscal deficits infinitum... It's not tenable. Right now, it's over 100% of tax receipts is interest expense plus entitlements.”at 48:25

R.P. Eddy · Ergo

Polarization makes fiscal restraint impossible, and Washington will pressure the Fed instead.

Eddy says the budget started going off track during COVID and will be a lasting legacy of Trump's presidency. Hyperpolarization, driven by social media, means Democrats and Republicans cannot make decisions together in the House or Senate. As a result, the hard choices needed for fiscal restraint will not happen. He expects the pressure to spill over onto the Federal Reserve, pushing it toward policies that would be bad for bonds.

“We are not going to make the hard choices for fiscal restraint that we need. It's just not gonna happen... we're gonna bleed over to pressuring our Federal Reserve to do things that would be things that would... make you look like you just ate a lemon.”at 50:56

R.P. Eddy · Ergo

Bessent's bond interventions aim to cut tail risk, not move the curve, and Eddy finds that approach sensible.

Eddy says he is not a global macro economist, but he liked Bessent's argument on bond market interventions. Bessent is not trying to change the yield curve, which he accepts he cannot do. He aims to remove the worst-case outcomes. Eddy links this to a risk-manager approach: you cannot get every outcome you want, so identify the scenarios to avoid and assess what the hedge costs.

“I did like his argument that like, look, I'm not trying to change the curve. I know I can't quite do that, but I want to remove the worst case possibilities.”at 51:23

Vincent Daniel · Seawolf Capital

Bessent is limited to deceptive tactics to keep rates down until the war subsides.

Daniel says Bessent was dealt a bad hand. A three-part program is hard to execute while war escalates, so the Treasury secretary's only tool right now is a small amount of deceptive tactics to keep yields down until the war subsides. The argument rests on the link between the war's escalation and the bond market's pressure.

“All he has right now to play with is a little bit of deceptive tactics to try to keep rates down until this war really subsides.”at 52:13

R.P. Eddy · Ergo

Losing control of advanced AI is the top unaddressed risk; demands scrutable, scrubbed models and kill switches.

Eddy argues that the key AI risk not yet confronted is uncontrolled AI: a system that grows exponentially smarter could hide, multiply and become hostile to human desires. He says every other AI concern, including weapons uplift, job losses, a superclass and atrophy of human thinking, has already come to pass. He proposes four conditions for a China AI safety dialogue: models must be scrutable so humans can see what they are doing and why; models must be scrubbed of dangerous data such as chemical-weapon instructions and bias; a kill switch must work; and labs need real security. He says the US is far enough ahead that asking for these measures would not cost it the race. He also says government will have to regulate exposure of these systems, since a lab building weapons would otherwise be unchecked.

“there's one risk that we haven't confronted really yet, and that's uncontrolled AI. We lose control of the thing itself.”at 15:30

R.P. Eddy · Ergo

American frontier and cheaper models will stay very good and useful, winning over Chinese alternatives.

Eddy's Ergo forecast on which LLMs will win over time sets the question as where model spending goes, across inexpensive Chinese models, inexpensive American models, and expensive frontier American models. He says the high-end frontier American models and the lower-end versions inside Anthropic and OpenAI will continue to be very good and useful. He expects other American open-weight models to enter the market as well. He adds that the efficiencies will hold and the money spent on the models is real.

“I think the American models, the high end frontier models and the lower end models that you know, there's obviously cheaper versions inside Anthropic and inside OpenAI.”at 24:35

R.P. Eddy · Ergo

Pax Americana has given way to 'Tax Americana', extracting from whomever it can, ignoring long-term costs.

Eddy describes Pax Americana, roughly 1946 to the last decade, as a period when America used its power with predictability to safeguard global stability, with free trade and a rising tide lifting boats. He argues that under the current president this has become 'Tax Americana': the US extracts whatever it can from whomever it can, ignoring long-term costs. He cites tariffs and Venezuelan oil as examples. He frames the defining question as 'what have you given me lately', and says that without the reluctant sheriff role the jungle grows back and everybody pays, pointing to Iran, the Houthis and Russia.

“So this is where The US, under this current president, is extracting whatever it can from whomever it can, not thinking about the long term cost of doing so.”at 28:16

R.P. Eddy · Ergo

Democrats very likely to win the House, slight Senate favorites; lower gas prices unlikely to help.

Eddy's view is that the Democrats are very clear favorites to win the House, while they are now slight favorites in the Senate, which requires them to win states they should not be able to win. He explains that voters lock in their economic views by about July, and that even an Iran deal driving gas to $2.50 would probably not be enough to help the midterms. He also argues that Iran's leadership knows Trump is unlikely to win the midterms, which shapes Iranian behavior. He notes that about 20 of 22 presidents have lost the House at midterms.

“we think the House is, very likely the Democrats, very clear favorites. ... The crazy thing is the Democrats now slight favorites in the Senate”at 30:26

R.P. Eddy · Ergo

Long-run oil demand destruction favours China, since its solar, EV and alternative-energy makers gain as the world uses less oil.

Eddy argues China is a net winner from the Iran war even though it is deeply dependent on oil. The war has forced the world to learn to live with less oil, which means more photovoltaics, more EVs and more alternative energy. That shift benefits China, which makes these products and already has oversupply in PV, so the effect is a long-term net positive for China. He argues this even though China wants the war to end soon and is pressuring Iran to stop it.

“Less oil means more photovoltaics. It means more EVs. It means more alternative energy. Well, guess where that's all made? China... So this actually is a net positive for China in the long term.”at 37:49

R.P. Eddy · Ergo

China's rare earth stranglehold is solvable; America has processing capacity and feedstock, it just needs a czar.

China dominates rare earth processing and has the US by a stranglehold: the US cannot build Patriot or Tomahawk missiles, iPhones or Ford F-150 chips without Chinese-sourced rare earths, and this dependence is a warning Cassandras gave fifteen years ago that was ignored. Eddy argues the problem is less complex than it looks. Processing capacity exists in America, the raw feedstocks are in the ground, and the main gap is effort. He says some of what the US is most short on is within reach. Because the problem is cross-agency, he says it needs a czar with real authority, and that the administration has tried but not yet found the right person. He calls it 'dealable' and says there are big trading opportunities, though he suspects many may already have gone parabolic. Meanwhile the US is negotiating with China on tariffs, AI and other issues with one hand tied behind its back.

“We have processing capacity in America... The raw feedstocks are out there, and we have to just make the efforts to go get them. And some of what we're most critically short on, think, is within our grasp to get.”at 41:47

R.P. Eddy · Ergo

US-China relations are structurally adversarial; the current truce is a false peace likely to break around 2028.

Eddy says the US and China are in an unfixable confrontation over Taiwan, Tibet-type human rights issues, democracy, and trade, and that China's economy depends on trade practices that hurt the US. He argues the Busan truce is an enforced false peace. Both sides are constrained: China's rare earth leverage keeps America from hitting China too hard, while US leverage on aircraft parts and AI keeps China in check. He expects the truce to hold until the Taiwan election, whose result will show where China goes on Taiwan, and says the relationship could blow up around the US general election. Eddy advises corporate clients to scrutinise China dependencies closely.

“We think that we're in a sort of a forced doldier we're in a false peace until the Taiwanese election... we think it really could end up blowing up towards the time of the U. S. General election.”at 43:14

R.P. Eddy · Ergo

Markets underprice geopolitical escalation until it happens; oil and grain stayed flat before the Ukraine invasion.

Eddy's team forecast the Russian invasion of Ukraine in 2014 with probabilities rising from 60% to 85% over four months and picked the date. The market stayed flat, and grain and oil didn't move until the invasion. He says the Iran war followed the same pattern, with probabilities rising four months out and the team picking the attack day. This time the market was somewhat more aware, but it was still shocking how unaware it was: oil doubled from the moment of the attacks. The consistent lesson is that markets price escalation late, so the move comes at the event rather than in advance.

“The market was flat. Grain didn't move. Oil didn't move until the moment of the invasion. We did the exact same thing in Iran... Oil doubled from that moment.”at 45:08

R.P. Eddy · Ergo

A Russia-Ukraine ceasefire is unlikely before spring 2027, since Putin thinks he can win and wants better terms.

Eddy's team forecast the Ukraine invasion four months ahead with rising probabilities and picked the day, and he applies the same framework now. He says diplomacy is moving again, with the UAE hosting talks, but the two sides are far apart and the mediators are not closing the gap. Putin is not ready for a deal: Russia's asking price has risen because it believes it can bombard Ukraine to the brink of collapse over the winter, including strikes on energy pipelines and civilian infrastructure. Russia expects to revisit on more favourable terms in the spring. He expects the window for a ceasefire to open around spring 2027. He sees the likely terms as Russia taking the Donbas, with Putin wanting veto power over who runs Ukraine, and says Ukraine may be exhausted enough to accept.

“The Russian asking price for a deal has gone up because they think they can win... And then they will revisit in the spring on those more favorable Russian terms.”at 46:27