Double Feature: Durable Goods Orders & Industrial Production
What is often mistaken as broad resilience in business investment over the past year has actually been a heavy concentration of spending in the AI and broader high‑tech build out. It is not so much a rising tide that lifts all boats as it is condensed activity in one sector that masks a struggle in industries not directly tied to the tech build-out. Today’s data point to an encouraging growth pick-up outside of high-tech.
That dynamic is clearly visible in the durable goods data: orders for computers & related products rose 13.7% over the past year, running at nearly four times the pace of underlying core capital goods.
The same concentration shows up on the production side. Domestic output of high‑tech goods—including computers, communications equipment, and semiconductors—is up nearly 9% through January, while non-energy production excluding these components is up just 2.2%. Even trade flows tell the same story where growth in U.S. imports last year was largely driven by high‑tech equipment, with non‑tech import categories losing momentum.
We don’t see the AI‑driven investment cycle fading anytime soon, and capital flows into that space are likely to continue.
This morning’s data show there are early signs that more traditional areas of capex are beginning to firm as well. Core capital goods orders are stabilizing and output excluding high‑tech is starting to improve. Recent tax incentives in the One Big Beautiful Bill Act are supportive of broader investment, and the early‑year pickup in commercial and industrial lending suggests firms are becoming incrementally more willing to finance capex beyond AI. (…)
Orders excluding transportation have firmed in recent months and rose 0.9% in January which was three-times the consensus expectation.
Orders for core capital goods (nondefense ex. air) looked better too, up 0.6% in December (twice the 0.3% expectation), and that better-than-expected reading came on the heels of an upward revision. Shipments of core capital goods rose 0.9% which also handily exceeded expectations, but remains consistent with our expectation for real equipment investment to rise at a near 4% annualized clip in Q4 when data print Friday.
Industrial production picked up at the start of the year, rising 0.7% in January, though downward revisions to year-end take some of the shine off the data. Output was supported by firmer utilities output, but most of the strength stems from a sizable pickup in manufacturing output.
Total manufacturing production rose 0.6% in January, posting the strongest monthly gain in nearly a year, and activity looks fairly broad based under the headline. Durable goods production was responsible for most of the strength (+0.8%) with all but the aerospace industry posting a rise in output. Nondurables industries were more mixed (+0.4%), with four of eight industries seeing output fall and a rebound in chemical output and some others offsetting declines.
You do not need to do a deep dive to find the theme of a tech-concentration in the industrial production data. Just look at the index levels. The Federal Reserve sets them all equal to 100 in 2017. The manufacturing index in January came in at 97.5 meaning manufacturing output, in volume terms, is 2.5% smaller than it was almost a decade ago. Yet the category “selected high-technology industries” came in at 184.7 the same month. This grouping includes output of computers, communications equipment, semiconductors and related electronic components.
While high-tech continues to outpace non-tech focused output, we see signs of recovery. Excluding high-tech, manufacturing output rose 0.6% in January, marking the fastest monthly increase in nearly a year with this measure of production matching its highest index reading of the past two and a half years.
The fact remains that overall manufacturing production is very spotty with no clear upward trend:
From BofA:
Data suggests that US new motor vehicle (“auto”) sales began 2026 on a soft note. On a seasonally-adjusted annualized rate (SAAR), sales fell to 15.4 million in January, from 16.4 million in December 2025, according to data from the Bureau of Economic Analysis (BEA). (…) auto sales have struggled to return to their 2019 level over the past six years, apart from an immediate post-pandemic rebound and a brief surge in early 2025, when sales rose in anticipation of higher prices due to tariffs.
We think affordability is a key factor likely driving weaker auto sales. Car prices have risen significantly over the past six years, though that rise appears to have largely stopped for now. New car prices are up around 22% on 2019 levels, while used car prices have jumped around 30%, according to data from the Bureau of Labor Statistics (BLS). At the same time, the cost of insuring a vehicle has risen significantly. When combined, these hikes may have curbed vehicle demand, especially compared with other areas of consumer discretionary spending. (…)
Exhibit 5 shows that the average auto loan payment from Bank of America customer accounts has risen most for younger Millennials (ages 30-36), with the average auto loan payment from this cohort up nearly 60% from 2019. Older Millennials and Gen Z have also experienced sizeable increases, each exceeding 40%.
In some respects, it makes sense that younger Millennials have seen the largest rises in average loan payments. In the early 2020s, many were likely engaged in household formation and potentially having children, all of which can precipitate the need for a new or used – often larger – car. And they were doing this at a time when car prices were elevated and auto loan financing rates had risen as a result of Federal Reserve rate hikes.
Fed Minutes Reveal Little Appetite for Rate Cuts Some officials favored more neutral language pushing back against the prospect of future rate cuts
Federal Reserve officials signaled little appetite for reducing interest rates at their meeting last month, with most indicating they wanted to see further progress on inflation before considering any more cuts—a process that could take months.
Moreover, even though two officials opposed the decision to hold rates steady at their Jan. 27-28 meeting and favored a cut, minutes of the meeting showed that other officials would have supported more neutral language characterizing the prospect of a rate cut or a rate increase as evenly balanced.
The minutes said those officials would have been comfortable changing the Fed’s carefully drafted postmeeting statement to reflect the possibility that rate increases could be warranted if inflation continued to run above the Fed’s target.
“Several participants indicated that they would have supported a two-sided description of the committee’s future interest rate decisions, reflecting the possibility that upward adjustments” in interest rates “could be appropriate if inflation remains at above-target levels,” said the minutes. (…)
The minutes, released Wednesday with a customary three-week lag, revealed that more officials were less worried about the labor market than they had been and more apprehensive about inflation.
Most officials, the minutes said, cautioned that progress bringing inflation down “might be slower and more uneven than generally expected.” The risk that inflation might run persistently above the Fed’s 2% goal “was meaningful,” they said. Likewise, the Fed’s staff inflation forecast described an interval of more persistent, above-target inflation as “a salient risk,” according to the minutes.
The minutes laid bare a lingering divide over where to set the bar for lowering rates. Several officials said they would still be open to cutting so long as inflation declined in line with their expectations. But a somewhat larger group said they had a higher bar that called for a clear indication that inflation progress “was firmly back on track” before cutting. (…)
This last chart plots my “Inflation on Essentials” series (CPI-Food + Energy + Shelter) in YoY (dash-left) and absolute vs Headline CPI. Current “Inflation on Essentials” costs are 30.3% higher than in Jan. 2020 while total CPI is 26.3% higher. YoY inflation is +2.7% on Essentials vs +2.4% overall.
The Atlanta Fed just released its Business Inflation Expectations survey:
Firms’ year-ahead unit cost expectations decreased to 1.9 percent.
Firms reported a median 3.0 percent (3.7 percent mean) price increase over the past 12 months and a median 3.0 percent (3.1 percent) expected price increase over the next 12 months. Realized price and expected price both increased from November (3.0 percent median realized price increase, 3.0 percent median expected price increase).
- Overall, firms are optimistic on sales and margins:
- They expect to raise their prices by 3.1% this year vs overall inflation of 1.9%.
Axios:
The Supreme Court could rule on the legality of President Trump’s tariffs as soon as tomorrow, and a raft of analysis will surely follow as economists and Wall Street researchers parse the fallout.
Just yesterday, Trump economic adviser Kevin Hassett called a paper from researchers at the New York Fed “an embarrassment” and said its authors should be “disciplined.”
The research found that American consumers and companies paid about 90% of the cost of Trump’s tariffs last year following several other reports with similar conclusions.
The dustup was a reminder of the trigger-happy political atmosphere around economic research, heightening concerns that Trump’s criticism is muffling commentary from Wall Street.
- Multiple firms have been toning down and censoring their research for fear of angering the White House, Bloomberg reported.
In August, Goldman Sachs chief economist Jan Hatzius put out a report also saying Americans pay some portion of the tariff cost. Trump then posted the bank’s CEO Davis Solomon should fire the well-respected analyst.
- Earlier this year, Treasury Secretary Scott Bessent said the head of Deutsche Bank called him to say the bank “does not stand by” an analyst report that suggested European investors may step up sales of U.S. assets. The bank later said the independence of its research was “sacrosanct.”
- And last April, a JPMorgan analyst voluntarily redacted his analysis of White House policy, literally blacking out portions of a report.
Asked whether the White House comments chill the speech of economists and analysts, spokesman Kush Desai said:
- “What should chill the work of these economists is the fact that they have consistently beclowned themselves by pushing one doom-and-gloom prediction and ‘study’ after another about President Trump’s tariffs.”
- Trump, Desai said, “implemented a historic tariffs agenda while cooling inflation, renegotiating trade deals, accelerating economic growth and securing trillions in investments — the exact opposite of what these economists predicted would happen.”
Wall Street research isn’t like a policy white paper or even an academic report. It’s intended for investors.
- “I don’t think people in Washington understand what our research reports are for,” one person at a top Wall Street bank tells Axios. “They’re written for clients.”
- Banks have a fiduciary duty to their customers. They can’t really change their findings to suit the administration.
- This person’s advice to bank analysts: Be clinical, not colorful and do good work for your clients and you’ll “be fine.”
“I try not to be political, but on a scientific question or an economic question, I’m going to state exactly my mind,” says David Kelly, chief global strategist at JPMorgan Asset Management.
- Analysts say what they think, he says, and those that don’t have “no business doing the job.”
The findings of the New York Fed report that Hassett criticized are not controversial among economists both conservative and liberal.
- “I think the findings of the paper are consistent with what standard economic analysis would suggest,” Michael Strain, an economist at the conservative American Enterprise Institute, tells Axios.
- “I trust the Fed research,” Kelly says. “I actually came up with the same result.”
The Supreme Court could issue its tariff ruling tomorrow or next Tuesday or Wednesday.
- As of this morning, there was a 26% chance the justices uphold the tariffs, per Polymarket.
GOP angst over voter turnout builds as losses pile up
Republicans are getting crushed in scores of state and local races, raising deep concerns about a deflated base refusing to show up to vote even in the most pro-Trump areas.
The numbers are startling. In race after race, Democrats are outpacing their 2024 performance by double digits, a clear sign of a yawning enthusiasm gap.
Democrats have outperformed former Vice President Harris’ 2024 numbers by an average of 10.5 percentage points in the 20 state legislative districts that’ve held special elections this year.
- Democratic candidates outperformed Harris by even more — an average of 13.9 points — in the 67 state House and Senate races last year, according to The Downballot, a site that tracks state-level and congressional campaigns.
Republicans’ internal polling is aligning with recent surveys that suggest a downturn in support for GOP candidates.
- Many Republicans trace their troubles to Trump not being on the ballot this year or in 2028.
- But strategists acknowledge that some of his actions — including the administration’s reluctance to release more of the Epstein files — have turned off parts of his MAGA base, while energizing Democrats and anti-GOP independents.
- Polls have shown widespread dissatisfaction with Trump’s immigration crackdown and with how he’s handling the economy.
A big warning sign came Jan. 31, when Democrats snatched a North Texas-based seat in the state Senate. Democrat Taylor Rehmet won the seat by 14 points in a district Trump won by 17 in 2024.
Then, on Feb. 7, Democrat Chasity Verret Martinez won a South Louisiana state House district with a 24-point landslide margin. Trump had won the district by 13 points in 2024.
On Feb. 10, a Republican won a special election for a conservative north-central Oklahoma state House district by 28 points. Trump won the district by 58 points in 2024.
Since the start of the year, Republicans have suffered double-digit drop-offs from Trump’s 2024 performance in state legislative elections in Northern and Central Virginia, New York City, east-central Minnesota, and southeastern Connecticut.
- “While it is tempting for many in our party to wish away these results,” a GOP operative told Axios, “the pattern is clear that there is at least a current 10-point Democratic over-performance from Trump 2024 — and it’s built on a fired-up Democratic base and a sleepy GOP base.” (…)
Some Republicans caution against reading too much into state and local elections, arguing they often don’t reflect national trends.
- They note that Trump’s cash-flush political operation didn’t aggressively work to turn out the president’s supporters in any of the recent elections — something it’ll do in U.S. House and Senate elections this November.
- They also point out that Trump plans to hit the trail aggressively, which they believe will help to turn out his supporters.
“Let’s not pretend a couple of low-turnout special elections suddenly signal a political earthquake,” said Mason Di Palma, communications director for the Republican State Leadership Committee. “They are unique, low-turnout contests driven by highly localized factors.”
AI CORNER
Pentagon-Anthropic battle pushes other AI labs into major dilemma
As the Pentagon and Anthropic wage an ugly and potentially costly battle, three other leading AI labs are also negotiating with the department — and deliberating internally — about the terms under which they’ll let the military use their models.
Defense Secretary Pete Hegseth wants to integrate AI into everything the military does more quickly and effectively than adversaries like China. He’s insisting AI firms give unrestricted access to their models with no questions asked — and showing he’s willing to play hardball to force their hands.
The Pentagon is threatening to sever its contract with Anthropic and declare the company a “supply chain risk” because it’s unwilling to lift certain restrictions on its model, Claude.
- The company is particularly concerned about Claude being used for mass domestic surveillance or to develop fully autonomous weapons.
- The use of Claude in the Nicolás Maduro raid deepened tensions. The Pentagon claims an Anthropic executive raised concerns after the operation, though Anthropic denies that.
- Administration officials say it’s unworkable for the military to have to litigate individual use-cases with Anthropic before or after the fact. “We’re dead serious,” a senior Pentagon official told Axios of the threat to cut off Anthropic and force its vendors to follow suit.
Crucially, Claude is the only model available in the military’s classified systems through Anthropic’s partnership with Palantir.
- Three other models — OpenAI’s ChatGPT, Google’s Gemini and xAI’s Grok — are available in unclassified systems, and have lifted their ordinary safeguards as part of those agreements.
- Negotiations to bring those companies into the classified domain are now more urgent as the Pentagon ponders how to replace Claude if necessary — a process a senior official conceded would be massively disruptive.
- Anthropic says it remains committed to working with the Pentagon, despite the public feud, and both sides say they might still come to an agreement.
- One acknowledged that the fight with Anthropic was a useful way to set the tone for negotiations with the other three.
Officials are adamant they won’t budge on a standard allowing the Pentagon “all lawful use” of the AI models, and a senior administration official said one of the three labs already told the Pentagon it was “ok with ‘all lawful use’ at any classification level.”
- A source familiar told Axios that it was xAI, whose founder Elon Musk has ripped rivals like Anthropic and OpenAI as “woke” for their approaches to safety. xAI did not respond to multiple requests for comment.
- Notably, xAI was the only bidder out of the frontier labs in the Pentagon’s autonomous drone software contest.
- OpenAI is bidding in a limited way to translate voice commands into digital instructions, but not for drone control, weapon integration, or target selection.
The senior official said the administration was confident the other two labs would agree to “all lawful use” across both domains. But sources familiar with those dynamics tell Axios it’s not nearly that clear-cut.
- An OpenAI spokesperson told Axios that moving into classified work “would require us to agree to a new or modified agreement.” Google declined to comment.
- The “all lawful use” requirement is hardly relevant for unclassified work. “People are going to use this thing to make their PowerPoint slides a little bit more quickly and easier. They’re not going to be developing autonomous weapons,” one source said.
- But applying that standard in the classified domain poses thorny ethical dilemmas.
While Anthropic CEO Dario Amodei has been the most vocal about the risks of advanced AI, executives at OpenAI and Google share some concerns about how their models might be used, sources familiar with those dynamics say.
- The companies may also fear revolts among their engineers, like the one Google experienced in 2018 over a previous initiative, Project Maven, that involved using AI to analyze drone footage. Google walked away from that deal after a damaging internal fight.
Then there’s the matter of how you ensure the Pentagon is complying with whatever usage terms have been agreed, or even with the law.
- “The whole game” is building infrastructure that ensures what’s being deployed is safe, and having oversight on the back end into how it was used, one source said.
- The source was skeptical of Anthropic’s claim that the company has sufficient visibility into Pentagon operations to ensure it’s comfortable with every use of its model.
- A separate source said Anthropic does have visibility and is confident its usage policies are followed.
One source familiar with the ongoing discussions said one issue is that the companies themselves don’t fully understand how their models will respond in certain scenarios, or why.
- “That is more challenging than just figuring out like, ‘Hey, will this metal withstand this degree of heat or that degree of heat?'”
- The source added: “If there’s a one in a million chance that the model might do something unpredictable, is that one in a million chance so catastrophic that it’s not worth taking a 1 million chance?”
If an AI model enables an autonomous weapon to near-instantly take down dangerous drone swarms, is it ethical to deploy it when there’s some small chance it could also fire on a civilian flight?
- Those are the sorts of questions the labs are grappling with.
The Pentagon’s position is that such decisions should be made by the military, not by executives in Silicon Valley. Anthropic and its rivals are under pressure to decide whether they can live with that.
How’s the Rule of Law in the USA these days?
There’s also the Rule of Money with Elon Musk putting pressure on everybody.
Shouldn’t this debate involve other world instances, countries?
BTW: The U.S. is assembling the most air power in the Middle East since the 2003 invasion of Iraq, The Wall Street Journal reports.
- According to Article I, Section 8 of the U.S. Constitution, only Congress has the explicit power to formally “declare war”
- Instead of formal declarations, modern conflicts (like those in Iraq and Afghanistan) are typically conducted via an Authorization for Use of Military Force (AUMF) passed by Congress.
- The President is widely recognized to have the authority to respond to sudden attacks on the U.S. or its interests without prior congressional approval.
In a social-media post on Wednesday, Trump said the fleet of US ships he’d ordered to the region, led by the USS Abraham Lincoln aircraft carrier, is “ready, willing, and able to rapidly fulfill its mission, with speed and violence, if necessary.”
Via John Authers:
Prediction market bettors now put the odds of a US attack at around 70%, while crude oil prices are rising. Brent surged 4.3% on Wednesday to top $70 once more:
Crude oil and gasoline prices troughed in the US in December. Oil is up 15% and gas 8% since.
