Crypto News: Fed now matters less than AI and Trump

Today, as the Federal Reserve emerges from its latest policy meeting with a well-anticipated stay-put stance, there's plenty to talk about in financial markets. But most of it isn't driven by the central bank. It's almost as if the Fed is an afterthought given the flurry of activity out of the Trump Administration and the DeepSeek AI chatbot out of China, which has dashed some of the belief that big US companies will come to dominate the new technology.

If I had to sum up my analysis of the incoming information, I'd say it has added heavily to downside risk — on inflation, on the chances of an AI bubble bursting, and even on the odds of a recession. So let's unpack it in the context of the Fed's decision Wednesday and what that means for financial markets going forward.

My train of thought is as follows:

The base case from last week was that inflation is more of a problem than the market thinks. That's still the case.
This explains the Fed's decision to hold the fed funds rate steady. Because that move was largely expected by markets, it wasn't a big driver. What markets want to know now is how long we'll be in this holding pattern and hawkish hints from the Fed did impact the market.
If I'm right and inflation will remain sticky above 2 1/2%, then tariffs and deportations will make doubly sure of that, meaning the Fed will be on hold for a long time. That will keep interest rates elevated with even the outside chance of 6% long bond yields.
Still, the bigger risk is the bursting of the AI bubble because DeepSeek is effectively a commoditization of the AI chatbot, threatening the seemingly endless supply of money being shoveled into AI development.
Downside risk in both stocks and bonds will be with us for a long time because of the fiscal, governance, geopolitical, inflation and AI overhangs. We can't even rule out a recession, which would mean a need to re-balance portfolios aggressively.

The Fed matters a lot less now

Given all of the macro risks playing out now, a quarter-point move here and a policy hold there by the Federal Reserve really doesn't matter. Not only was this hold expected, markets expect the Fed to be in this holding pattern for the indefinite future. The first real chance of a rate cut is in June or July according to swaps markets. And even that is speculative given the Fed's lack of certainty about inflation coming back to target, a core inflation number expected to come out at 2.8% this Friday, and an economy which the Atlanta Fed's popular GDPNow economic tracker says grew at around 2.3% last quarter.

I'd go almost as far as saying what the Fed does now is irrelevant. Sure, you've heard the likes of Fed Governor Waller, Chicago Fed President Goolsbee and New York Fed President Williams all talking about the potential for more rate cuts. The reality though, given the data, the Fed was forced to remove any reference of inflation making progress toward its 2% goal in the policy statement. In essence, we won't get cuts until inflation comes down further, the jobs picture weakens or the impact of the Trump Administration's stated policy goals become evident. Of those three items, I'd watch the Trump part most.

Trump's unpredictability adds little upside in a market priced to perfection

Given where we are, with the bond market priced for more rate cuts and stocks priced for only the best of times, there's almost no upside one could imagine from fiscal and foreign policy. None. I mean the economy is already growing well above trend with elevated inflation. Anything Trump does to boost growth further is more likely to push inflation higher — and spur rate HIKE fears — than it is to be treated as something good for financial assets. And a lot of what I hear from Trump sounds like a recipe for recession.

Take the memo released late on Monday suspending hundreds of billions of dollars in federal grants and loans. Bloomberg reported that, "at least 20 states experienced problems accessing the Medicaid system… Oregon Senator Ron Wyden posted that his staff confirmed reports that Medicaid portals [were] down in all 50 states." Or take the reported offer to allow federal workers to work remotely — but only if they agree to leave their jobs. Both of those are bad for the private sector income.

The White House rescinded the grant freeze Wednesday, in a surprising reversal. But the flurry of actions is sowing chaos and creating the risk of a constitutional crisis over who, Congress or the President, has authority for these types of actions. A judge ended up blocking the order minutes before it was due to take effect. But from a markets perspective the message is clear: administratively, literally anything can happen in the US government now. We could see millions of federal employees without pay, millions of agriculture or building workers showing up less reliably for fear of immigration issues, and we could see tariffs on just about any country for any reason — as the spats with Denmark and Colombia show.

Donald Trump thinks this unpredictability is his ace in the hole for extracting gains from adversaries over the long term. But one of these episodes will result in a policy misstep from the US or abroad with unknown economic consequences. For a market that saw trillions of dollars in losses just because of the revelation of a Chinese AI chatbot, that's a lot of downside risk.

AI is in trouble now too

The AI market meltdown on Monday tells us a few things about how vulnerable asset markets are. But let's go into what the actual DeepSeek chatbot episode tells us about artificial intelligence before we talk about consequences.

I'd sum up Deep Seek this way: A Chinese company on a veritable shoestring budget delivered a product basically as good as ones that resulted from hundreds of billions of dollars of collective spending on artificial intelligence.

If DeepSeek can achieve masterful results on a limited budget, it suggests that a lot of the money already spent on AI has been wasted. The conceit was that bleeding-edge hardware and reams of computing power and data were needed to deliver best-in-class results. But DeepSeek shows us that simply isn't true.

When I asked ChatGPT "how much money has been spent on artificial intelligence," after spitting out some data points, the chatbot summed it up this way:

The global AI market (encompassing both hardware and software, services, and applications) was valued at $136 billion in 2022 and is projected to grow at a compound annual growth rate (CAGR) of around 38% over the next decade, potentially exceeding $1.8 trillion by 2030.
In short, while it's difficult to account for every dollar, it's safe to say that the global AI industry is already a multi-trillion-dollar market and is rapidly expanding with both private and public sector investments accelerating.

That's a lot of money. And given the doubts we should now have about how much money one needs to best create a gold-standard large language model, a lot of it is probably wasted. And that means those projections of future spend are going to have to come way down.

Hence the meltdown in artificial intelligence stocks led by Nvidia and of energy companies, too, who were banking on the need for gobs of energy from the big tech companies for AI uses.

Think of ChatGPT in 2025 as AltaVista in the year 2000

I think there are a lot of parallels to the Internet Bubble here. We have this new technological advance which will have huge ramifications for the future. But no one knows how much money to invest, how long it will take before investments garner profits, or what business models will make any money. Least of all, we don't know who will come out on top.

It's just a 'build it and they will come' Field of Dreams story at this point. Every single one of the companies we see now active in artificial intelligence could end up being a market loser 10 years from now.

In the late 1990s, there were a host of search engines trying to take what seemed at the time to be firehose amount of data and help users navigate to exactly the information they were looking for. Companies like Lycos, Excite, Yahoo, InfoSeek, AskJeeves and AltaVista were all plying their trade well before the dominant player today, Google, was founded in 1998.

If you had asked anyone who would be the winner, AltaVista would have been as good an answer as any. I asked ChatGPT "what happened to AltaVista?" It said this:

"AltaVista was launched by Digital Equipment Corporation (DEC) in December 1995. It quickly became one of the most popular search engines of the mid-90s due to its advanced technology, which indexed the full text of web pages and returned faster and more relevant results compared to other engines at the time.
[...]
...AltaVista was acquired by Overture Services (an internet marketing company) in 2003, which itself was later purchased by Yahoo!. AltaVista was eventually folded into Yahoo!'s operations."

Translation: AltaVista was an Internet loser.

Search became a huge revenue generator because of advertising. But AltaVista wasn't the one who got the dollars. It was Google, a very late entrant into the crowded search market.

I think we see capital expenditure dwindle

So what's the lesson for us today? It's that these sky-high prices don't reflect reality. They are options. The stock prices are like a call option on future profits with huge amounts of implied volatility. That optionality is worth a lot of money – as we saw with Amazon and Google, so much that it creates the kind of price movements that naturally lead to a mania.

But make no mistake, most of these AI companies will be 'losers' in the AI land grab. And their share prices will fall accordingly. At least big tech companies have existing business to fall back on, like Meta did after its metaverse pivot flamed out spectacularly.

DeepSeek represents a commoditization of the building blocks of AI, not an end to AI Investment. But my thesis is the same as what I presented to you six months ago: Eventually "the big tech companies need to show profit growth emanating from investments in AI. If they don't their shares will get hit." By putting a question mark over the investments made to date, DeepSeek has moved us closer to that point. As a result, I believe we are now closer to a major pullback in AI capital expenditure.

Market implications of the Fed, Trump and AI are negative

We seem to have reset to a world with inflation above 2%, a world akin to the couple of decades we saw before the housing bubble popped. That makes it hard for the Fed to lower interest rates. In fact, if you look at the 1990s and the 2000s, a 5% fed funds rate or higher was the norm when the economy was doing well. It only ever dropped below that level due to calamity or the hint of recession.

With the 10-year bond yield at just over 4.5%, essentially the same level as the fed funds rate, we seem to be about as low as we can expect to go without expecting a recession and rate cuts. Any upside risk to inflation or employment will take us higher. Starting around 4.75%, so not that far away, the negative implications begin for stocks. And they increase the higher rates go because stocks are priced for a lower rate, higher growth environment.

If you look at non-bubble stock valuations, so 1991-1995 and 2002-2008, the S&P 500's price-earnings ratio was always lower than 20 times and as low as 15. Discounting future earnings, especially when they are very back-loaded into the distant future by 5.5 or 6% is a lot different than imaging a 4.5% discount rate. That alone cuts 20% off the value of stocks.

Then you have Trump, who seems to want to radically re-shape the federal government. It was quite revealing that, in her first outing as Trump's press secretary, when asked how organizations that rely on federal funding should make payroll, Karoline Leavitt said they should call Russ Vought to make a case. We're talking about hundreds of billions of dollars in grants and loans, with one guy responsible for saying yea or nay. That's a bottleneck to say the least. While this action has been stayed by the US judiciary, we should fully expect some eventual breakdown in the administration of federal funds to the private sector. With that breakdown and a coming AI capital expenditure pullback, a recession can't be ruled out.

On AI, the fact that an open source operator can achieve results at a fraction of the cost should eventually also help big tech's bottom line since AI investment is a cost for most of big tech. And so, the promise of the DeepSeek AI results are good for the likes of Amazon, Meta and Apple.

But commoditization, though great for consumers, is always bad for profits. It says that the moats around the business of any firm in the artificial intelligence are smaller than we thought. Big tech was operating under the assumption that massive capital investment was a barrier to entry. And that meant that once they figured out the revenue model, they could reap massive rewards. Not only is the revenue model still unknown, DeepSeek suggests pretty much any decent-sized tech company can enter this space and compete well.

That's a negative for big tech, especially to the degree that new upstarts chip away at their existing revenue models too — especially Google and Meta in advertising. In a world where 80% of S&P 500 gains came from just seven stocks in 2023 and then another 60-70% chunk in 2024, there's a lot riding on this outcome.

Quote of the week

"it is too early to abandon Mag7/AI theme especially with earnings this week. That said, I do think this accelerates the timing on when the market will want to see an ROI on AI investments"

JPMorgan Chase