The Fed Has AI’s Back | Dan Krausz on “Liquidity Trap Door” In AI Infrastructure and Why Policymakers Are Inclined to Support Financial Conditions

Anthropic

Across the shows: 79 speakers in other episodes, 95 bullish and 61 bearish calls, past 30 days

• Dan Krausz · Blue Door Asset ManagementUncertain

  Claude could gain pricing power over firms that replace workers with it, but this is a distant risk.

  He uses Claude for about 95% of his own process, which he describes as table stakes. If a company replaces its employees with ChatGPT or Claude, it is far more dependent on the model provider than on any one employee, since it cannot easily replace the workforce. If only a handful of model companies offer the service, they could take pricing and the customer would have little recourse. He calls this a risk far down the road, with many variables that could prevent it, so he does not take a strong view.

  “Claude's kind of got you cornered now. When they decide to take pricing, there's not much you can do”at 46:31

NVIDIA

Across the shows: 49 speakers in other episodes, 40 bullish and 14 bearish calls, past 30 days

• Dan Krausz · Blue Door Asset ManagementBearish

  Nvidia can support the AI buildout, but its balance-sheet capacity is limited and markets may not reward it.

  Dan thinks Nvidia, and suppliers like it, have a lot of power and should be willing to support the buildout. But that capacity is not unlimited. Each time they are forced to step in as lender of last resort, markets start asking how many more times they can do it. Even if Nvidia says it is there during tightening, he does not think that will be enough. He draws a parallel with 2025, when hyperscaler capex was funded from cash and attracted little scrutiny, but questions about sustainability began once free cash flow turned negative and borrowing started. He is not sure this support would be capitalised into the stock's multiple.

  “I think it should be high. I think it is there. I think they have a lot of power, but the problem with it is it's not unlimited.”at 1:15:07

AI infrastructure & data centers

• Dan Krausz · Blue Door Asset Management

  Policymakers are unlikely to deliberately derail the AI buildout, given how crucial they say AI is.

  He argues policymakers have repeatedly called AI crucial for both inflation and competitiveness, with Treasury Secretary Bessent saying winning the AI race is crucial and Fed Chair Warsh calling AI a major disinflationary force. Raising rates to stop AI would contradict that stance, so he expects policymakers to avoid excessive tightening for AI. He cites the Fed's absence of AI-related overheating talk as evidence they are aware but not targeting AI. He also notes commentators floating a Fannie Mae-style vehicle for AI financing; he says it is not happening tomorrow but that many options can be explored. Inflation is the main risk: if it stays hot, policymakers may be forced to tighten, which would hurt the trade.

  “And so I'm still going with the assumption that they will try to not create too much tightening for AI to be able to continue with the build out.”at 32:03

• Dan Krausz · Blue Door Asset Management

  AI capex will keep growing while liquidity funds it; demand isn't the problem, and spending may near peak rate of change.

  He argues AI capex is not peaking. Demand has stayed strong and there has been no demand problem yet. The real constraint is financing: the capex was funded by hyperscaler balance-sheet cash but is now funded by markets, so it depends on liquidity. He expects capex to keep growing as long as liquidity supports it, though the rate of change may be peaking. He says AI spend could slow in a more material way by the back half of next year if tight liquidity persists. He views companies holding capex guidance flat (no raise) as an appropriate and necessary pause, because raising capex into weaker liquidity makes it harder to finance.

  “Well, we might be peak rate of change or approaching peak rate of change. I don't think we're peak. I think it's going to continue to grow as long as the liquidity is there to service it. That's the key because the demand is there.”at 34:13

• Dan Krausz · Blue Door Asset Management

  AI buildout is liquidity-dependent; financing, not AI demand, is where the bear case plays out.

  Companies are finding real uses for AI, so demand is not the constraint. Whether the capex gets financed depends on liquidity. Without capital constraints the spending is likely to keep going and could become the bubble people fear. If liquidity tightens, the ramp becomes much slower. He says the AI infrastructure trade is liquidity-dependent, so weaker-liquidity infrastructure names make sense to sell during tightening. He is focused on the near term because liquidity weakening shortens the timeline on risks.

  “The financing is where the bear case will be playing out.”at 37:28

• Dan Krausz · Blue Door Asset Management

  Markets scrutinise hyperscaler capex once it is borrowed rather than funded from cash.

  Dan compares the current debate with 2025. Then, hyperscaler capex was funded from cash and drew little scrutiny. Once free cash flow turned negative and the hyperscalers had to borrow, investors began asking how long the spending could continue. He thinks the same dynamic will apply to Nvidia and anyone else funding the buildout from their balance sheets, which limits how much that support will lift market pricing.

  “in the same way that in 2025, you didn't think twice about CapEx from the hyperscalers because it was all coming out of cash. And then as soon as free cash flow goes negative and they have to borrow, suddenly people start talking about, well, how long can this go on for?”at 1:15:36

US Treasuries & yields

• Dan Krausz · Blue Door Asset Management

  Treasury will likely ease liquidity via front-end issuance and buybacks, since not doing so would be inconsistent.

  He says the Treasury is discussing several ways to increase liquidity, including issuing more at the front end versus the long end and buybacks. Issuing more at the front end provides liquidity and also accommodates the long-end issuance that AI borrowing needs. He is unsure whether they will follow through, but says it would be inconsistent for them not to. The administration has framed AI as existential for the AI race and as a disinflationary force, so it would be contradictory to let liquidity tightening damage the AI trade.

  “They're talking about issuing more at the front end versus the long end and all the buybacks and all that kind of stuff. These are all ways to increase liquidity.”at 10:02

• Dan Krausz · Blue Door Asset Management

  Yields reflect AI capex plus deficits above 8% of GDP, and over the medium term rate levels are a policy choice.

  Rates reflect the combination of AI spending and the fiscal deficit, which together are 8%-plus of GDP. Much of GDP is driven by areas that are hard to cut without tightening, so rates have adjusted upward. He argues that over the short and medium term, rates can be capped as a policy choice, at the cost of a weaker currency and other consequences. The Treasury could use twists, buybacks, or front-end issuance. He points to David Zervos's Treasury work and his 'don't fight the Treasury' view as evidence this is being considered.

  “over the short and medium term rates can be a policy choice”at 1:09:30

• Dan Krausz · Blue Door Asset Management

  Weaker growth next year could let long-term rates hang out lower, though strong nominal GDP, deficits and AI capex limit declines.

  Dan's baseline is that nominal growth of about 8%, the deficit and AI capex are all strong, and while they stay strong it is hard to get a large decline in rates. He does expect growth to be weaker next year, so he thinks it would not be crazy for rates to hang out somewhat lower if that happens. This makes him moderately bullish on rates, with the outlook conditional on growth slowing.

  “I do think growth is gonna be weaker next year, so it wouldn't be crazy for rates to kind of hang out a little bit lower if that happens.”at 1:11:42

US stock market outlook

• Dan Krausz · Blue Door Asset Management

  Equities are far more sensitive to liquidity than the economy, so markets act as an early warning signal.

  He argues that asset prices are much more sensitive to changes in liquidity than the economy is. A short liquidity squeeze can knock the S&P down 10%, which worries people and prompts policymakers to ease before the economy is hit. If tightening persists, the impact does reach the economy after six to nine months. This explains why markets have often priced recessions that never arrived: liquidity weakened, the market priced risk, and when liquidity returned, the economy was never meaningfully hurt. In his framework, the S&P is the tip of the spear that warns policymakers early.

  “is if liquidity stays tight long enough, it has a fundamental impact. And so what happens is the sensitivity of asset prices is much greater to changes in liquidity than the sensitivity of the economy.”at 16:46

• Dan Krausz · Blue Door Asset Management

  Equities should hold up without tighter liquidity, but a 10%+ correction is possible if it tightens.

  His base case is that equities are fine as long as liquidity does not tighten incrementally, because earnings should hold up and there is no large immediate threat. Incremental liquidity tightening would make a much bigger event possible, which he defines as a 10%+ correction. He thinks the market is more vulnerable to that now than it has been in quite a while. He does not see the upside story as over, because policymakers should keep providing liquidity while they still can.

  “as long as we don't get incrementally tighter liquidity, equities will be okay because earnings are going to be okay... if we get incremental tightness from liquidity, I think that we could have a much, much bigger event.”at 1:05:44

• Dan Krausz · Blue Door Asset Management

  If rates rise past previous highs, the equity market will face consequences.

  Dan sees strong deficits and AI spending as pushing rates up. He thinks that if rates go on to eclipse their previous highs, there will be consequences for the equity market. He says he is focused on the equity-market effect rather than whether that outcome would be good for the economy. He frames this as a short-to-medium-term issue.

  “But I do think there'll be consequences if rates go and eclipse the previous highs.”at 1:11:11

US rates & the Fed

• Dan Krausz · Blue Door Asset Management

  The Fed will keep hiking, but at a measured pace to avoid tightening financial conditions too much.

  He reads Warsh's October press conference as signalling more hikes, with a tightening bias, because inflation is too high to ignore. He says the Fed is trying to keep tightening financial conditions without tightening too fast, because a rapid tightening would be self-defeating and force a reversal. The measured pace preserves optionality. He notes the upgrade of economic language from solid to strong and the description of a hike as removing a dose of accommodation. He also highlights that Warsh said the economy is not overheating, which separates inflation from growth and differs from 1999 when Greenspan linked overheating to tech spending.

  “the Fed wants to continue to hike. They will continue to hike most likely, but they want to do it in a measured pace.”at 20:22

• Dan Krausz · Blue Door Asset Management

  Policymakers still have room to add liquidity, since the dollar is strong and they have not yet acted.

  He has long believed policymakers will keep providing liquidity for as long as they can, and that this is good for equities. The point where that stops working credibly would come after a substantial decline in the dollar, because adding too much liquidity would then be too dangerous. The dollar is strong, so there is plenty of room to add liquidity. Policymakers could also push banks to buy more Treasuries through regulation or incentives, and they have not done much of that yet. Liquidity could still tighten, but they are not yet trapped.

  “a decline in the dollar to where now it's too dangerous to increase liquidity too much... The dollar is strong. So, there is plenty of space for them to increase liquidity if they really wanted to.”at 1:06:47

• Dan Krausz · Blue Door Asset Management

  US policymakers lean toward supporting the AI buildout and capping long-rate rises unless inflation accelerates.

  Dan agrees that policymakers are on the margin inclined to support the AI buildout, growth, and to try to halt the rise in long-term rates. The qualification is that if inflation keeps accelerating, they will tighten and put credibility on inflation first, and will not act at any cost. If inflation is stable, he thinks the inclination is towards supporting the AI buildout.

  “But absent that, if inflation is stable, I think the inclination is towards supporting the AI build out”at 1:14:11

US deficit & government debt

• Dan Krausz · Blue Door Asset Management

  Large deficits add liquidity, especially when financed at the front end rather than the long end.

  He argues a larger fiscal deficit is liquidity-positive because the government effectively arbitrages its lower financing costs and places the funds into the private market instead of the private sector borrowing. The effect depends on how the deficit is financed: financing more at the front end is far more positive for liquidity than financing at the long end. This is why he sees the Treasury's front-end issuance choices as a key lever for liquidity. He also notes the 6% deficit is very stimulative, which makes it hard for rate hikes to cause broad weakness.

  “Fiscal deficit is positive for liquidity, but it also depends how it's financed. Financing it more at the front end is way more positive for liquidity than at the long end.”at 14:35

• Dan Krausz · Blue Door Asset Management

  Deficits have not yet trapped policymakers, since none of four stress signals is present, but the path is heading there.

  He watches four indicators to see when deficits turn negative for equities: a spike in term premium, weakening capital flows into the US, substantial dollar weakness, and rates above nominal GDP. None are present. Term premium has not spiked, capital flows into the US have increased, and the dollar has not weakened substantially. Rates are also not above nominal GDP. That makes this a classic liquidity issue rather than a trap. He says policymakers are on the road to being trapped given the size of the deficit if nothing is done.

  “So none of those are present right now, which means to me, this is a classic liquidity issue. ... We're on the road to the policymakers being trapped given the size of the deficit if nothing is done about it.”at 1:08:18

AI & jobs

• Dan Krausz · Blue Door Asset Management

  AI productivity will not show in economic statistics for a year or two; productivity data is nearly useless.

  Productivity is unobservable and heavily revised. Historically the biggest productivity gains show up after a recession, when output falls less than employment. Companies also need time to build out their AI use before they can lower costs by cutting staff, so the cost answer is still early. Company-level anecdotes will exist, but the economy-wide answer is not due yet. He calls productivity as a concept extremely important, but as a data point almost useless.

  “I don't think you'll see it in the statistics in one year or two.”at 56:25

• Dan Krausz · Blue Door Asset Management

  AI productivity may come at workers' expense, but this technology may differ from past ones that created jobs.

  Every past technological turning point raised fears that productivity would come at workers' expense, and the jobs usually came back. He thinks creating intelligence is different from the industrial revolution or railroads, so the concern may be justified this time. He is humble about that, since everyone at earlier turning points also believed their technology was different. He expects more data on employment effects over the next year, because the political lull lowers the chance of political distortion in hiring.

  “I do think that intelligence is something different, creating intelligence as opposed to industrial revolution and all that”at 1:00:01

Earnings & corporate profits

• Dan Krausz · Blue Door Asset Management

  Four structural profit tailwinds have ended, so labour productivity must now justify returns on capital.

  Over the last thirty to forty years, returns on capital and margins rose because of four trends: falling tax rates, falling interest rates, falling capital intensity of the economy, and falling labour share of GDP. He says the decline in tax rates is probably over, interest rates have clearly bottomed, and capital intensity has reversed with AI capex. The labour-share trend is what remains. Labour productivity therefore has to rise, measured as revenue per employee, both to help with the debt and deficit and to justify the AI capex.

  “Secular decline in interest rates clearly over. The secular decline in the CapEx intensity of the economy over. So labor becomes really important.”at 58:00

• Dan Krausz · Blue Door Asset Management

  AI-ecosystem earnings are exceptional but unsustainable beyond a year or two; revisions will weaken next year.

  AI-ecosystem earnings have been exceptionally strong, while earnings elsewhere have been fine rather than great. Strong growth concentrated in one area is normal for a boom, since others lag while still benefiting from a stronger economy. The AI-driven pace cannot persist more than a year or two. He expects earnings revisions next year to be weaker for the reasons he has covered, which will be a change because revisions have been very strong.

  “No, it's not sustainable to have that kind of earnings growth more than a year or two.”at 1:04:54

Credit markets

• Dan Krausz · Blue Door Asset Management

  Treasury and AI issuance get funded first, so any liquidity squeeze hits lower-quality credit hardest.

  He describes liquidity as a waterfall in which the largest and most important credits get funded first. Treasury issuance is the top bucket and AI issuance (the hyperscalers, investment-grade) is next, so both absorb capital before anything below. Because AI-related issuance has grown so much, the rest of the credit stack has less capital. Lower-rated credits are therefore hit first and hardest when liquidity weakens. Hyperscaler capex used to be cash-funded and so was insulated, but as it moves to markets it adds to the competition for liquidity.

  “the buckets on top are going get funded first. That's treasuries. The next buckets will get funded next. That's IG, like the hyperscalers.”at 2:17

Emerging markets

• Dan Krausz · Blue Door Asset Management

  EM currencies are typically hit first when liquidity weakens, and they topped in February this cycle.

  In past liquidity cycles, the asset that needs liquidity most is EM, so EM currencies get hit first when liquidity starts to weaken. This cycle has followed the pattern: EM currencies topped in February, even though the capex needs are now concentrated in AI rather than in the usual sectors. He presents this as evidence that tightening liquidity is already affecting the parts of the global system that need it most.

  “usually the thing that needs liquidity most is EM. And so EM currencies get hit first... EM currencies topped back in February.”at 3:53

US housing

• Dan Krausz · Blue Door Asset Management

  Rising rates keep pressuring housing, which has had only an L-shaped recovery.

  He describes housing as having had an L-shaped recovery rather than a rebound. Rising long-end and short-end rates continue to press on housing, and the rates-driven liquidity tightening hits the parts of the economy that are rate-sensitive. He groups this with small businesses that are okay but not strong and non-AI cyclicals that are growing but not particularly strong. He presents housing as one of the rate-sensitive segments of a two-speed economy in which AI capex is the main source of strength.

  “You've seen an L shaped recovery in housing. These rising rates just keep on putting pressure on housing.”at 7:13

Private credit

• Dan Krausz · Blue Door Asset Management

  Private credit lenders are tightening standards, reducing one source of lending liquidity.

  He identifies lending as a key source of liquidity, with credit standards affecting how much capital is available. Large-bank lending standards have been mostly well-behaved, though JP Morgan said on its earnings call that it was incrementally tightening. Private credit, another major lending source, has said it is tightening its standards. He argues this has an impact because less lending means less capital available, which adds to the liquidity squeeze that the market is pricing.

  “Private credit, another source of lending, they have certainly commented that they've been tightening their standards.”at 15:15

US growth & recession risk

• Dan Krausz · Blue Door Asset Management

  Growth will materially slow next year as tight liquidity and rates feed through, though not into recession.

  He expects growth to slow materially through next year, not immediately in Q1. He expects the ISM to weaken through the year and non-AI cyclicals to see weaker growth. Consumption is also likely to weaken. The lagged effects of the rate rise, including the long end as well as hikes, are still coming through. He says nominal growth will weaken, but he does not think it will be recessionary. Strong capex and a large fiscal deficit cushion the downturn, a point Jack Farley makes as well.

  “I expect we're going to get material slowing in growth next year as we go through the year.”at 21:18

US consumer

• Dan Krausz · Blue Door Asset Management

  Consumer spending will weaken next year as falling discretionary wallet share drags on consumer discretionary stocks.

  He tracks the wallet share available for discretionary spending, which rose for thirty years as falling rates and commodity prices made necessities cheaper. It has fallen materially over the last four years and is a good leading indicator of spending. The year-over-year change has weakened substantially this year, which points to weaker consumption next year. He has been structurally bearish on consumer discretionary relative to the market: rates that are flat or higher now drag on relative multiples, and higher rates take a long time to feed into wallet share. A bounce is possible, but the structural decline continues even if oil falls to $60.

  “I think consumption is likely to weaken next year for a host of reasons. But I've been kind of a structurally bearish on consumer discretionary relative to the overall market.”at 21:40

Enterprise AI adoption & AI software

• Dan Krausz · Blue Door Asset Management

  Companies are finding real uses for AI; the early bear case on AI usage is not playing out.

  Two years ago the bear case was that AI would not be used as expected. He says that is not happening. Companies are integrating AI more, and the main thing most companies want to spend on is using AI to become more efficient. The question about demand has largely been settled in AI's favour. The open question is whether capital is available to fund it, which he treats as a liquidity question.

  “I don't think that bear case is playing out.”at 37:25

Market sentiment, positioning & volatility

• Dan Krausz · Blue Door Asset Management

  High gross and low net positioning forced hedge funds to unwind, creating a mechanical rally outside AI.

  Positioning data showed gross exposure very high while net exposure stayed relatively low. That means people were pressing relative trades, long AI hardware and short perceived AI losers. When AI hardware sold off, those funds had to buy the other side, which drove a rally in non-AI stocks. The Goldman AI-at-risk index, which has underperformed for about two and a half years, has risen over 50% from its lows as a result. He says that rally was driven by covering shorts, not by better fundamentals.

  “What that tells you is people are really pressing their relative trades.”at 38:14

Software & SaaS

• Dan Krausz · Blue Door Asset Management

  AI disruption risk to software and services firms persists; their recent rally was a short squeeze.

  Stocks perceived as AI losers rallied as hedge funds covered shorts, not because their fundamentals improved. The rally came from positioning. The disruption trade is not liquidity-dependent, since it is driven by new businesses being started with AI and companies using AI internally. He expects disruption worries to return next year regardless of liquidity, and says the terminal value question for these businesses remains open.

  “It's just that they were over shorted. They got a short squeeze, but the disruption risk is still very much there.”at 40:50

AI pricing & unit economics

• Dan Krausz · Blue Door Asset Management

  AI's long-term margin boost for companies is questionable because it erodes competitive moats.

  AI gives an initial margin benefit because it makes businesses more efficient. Once everyone uses it, it becomes table stakes and widens competition. Margin structures usually rest on moats, and AI lowers them. Competitive industries such as retail tend to price to gain share, so efficiency gains get competed away. Historically, when products become commodities, long-term margin benefits rarely last. The shale revolution is his example: productivity gains went to consumers and hurt producers. Oligopolistic industries could still keep margins, and fiscal deficits are also flattering margins now.

  “AI is built on lowering moats and democratizing competition in almost anything.”at 43:47

Open-source AI models

• Dan Krausz · Blue Door Asset Management

  Cheap Chinese or open models could commoditise intelligence over time; policy could interrupt it.

  He thinks the scenario where a company outsources to cheaper Chinese models that sell tokens at a fraction of the cost is relatively likely at some point. Competition is how these things usually evolve. Policy could interrupt it, for example by banning certain open-source models. He hesitates to take a strong view because there are many variables, and he would need 70-plus things to go right first.

  “I do think that what you described is a relatively likely scenario at some point,”at 48:40

US jobs & labor market

• Dan Krausz · Blue Door Asset Management

  Employment should be solid over the next six to nine months, based on consumption and unit labour costs.

  His favourite long-term employment indicator is consumption growth minus unit labour cost growth. When labour costs grow faster than consumption, firms lay people off, and the reverse drives hiring. Over the last year, unit labour costs have been tame while consumption has been strong. Historically that would imply solid hiring. Lagged effects from the past year also support that. If employment comes in weaker despite this, that would show the labour market is breaking from historical relationships. Employment has not been strong despite high nominal GDP, though unemployment and jobless claims remain low.

  “employment should be pretty solid over the next six, nine months.”at 1:00:52

US dollar

• Dan Krausz · Blue Door Asset Management

  Forcing rates to stay or fall could cause consequences in currency markets.

  Dan argues that policymakers face a choice. If they try to keep rates where they are or push them lower while the deficit and AI capex are strong, this may have consequences in the currency market. He presents this as a possibility rather than a firm forecast, and frames it as a short-to-medium-term choice.

  “If they decide we're going to force rates to kind of stay where they are or go lower, there may be consequences in the currency market.”at 1:11:16

Gold & precious metals

• Dan Krausz · Blue Door Asset Management

  Gold is bullish if Treasury shifts issuance toward the front end; a break in that pattern would be the signal to watch.

  Dan says gold is the asset most highly correlated with changes in how Treasury issuance is distributed. When Treasury issues more at the front end relative to the long end, this tends to be very bullish for gold. Gold is also very sensitive to liquidity, so it has been an early tell on liquidity. He expects Treasury to increase short-end issuance relative to long-end issuance, which should be bullish for gold. He would watch Treasury's issuance mix as the key gold indicator and would buy gold if that correlation broke in his favour, though he holds no gold position now.

  “Gold is the highest correlated asset that I have found with changes in how treasury issuance is happening... when Treasury issues more at the front end relative to long end, tends to be very bullish for gold”at 1:12:26