AMU

December 11, 2025
Final Thoughts
Written by Michael Cowden
I attended the Heating, Air-conditioning & Refrigeration International’s (HARDI’s) annual conference earlier this week in Las Vegas. It was a great excuse to get out of the cold and snow in Chicago. And I applaud HARDI for attracting more than 2,200 attendees!
I was there for a meeting of HARDI’s Sheet Metal Council. I made a short presentation and moderated a panel whose members have a lot more industry knowledge than I do. And I appreciated them sharing their expertise, and the good questions and different perspectives from folks in the audience – a mix of companies in the steel, HVAC, and distribution.
And so, first, I want to say ‘thank you’ to HARDI for inviting me back again this year.
Geeking out on galv, HVAC, and trade
I’ve been participating in HARDI’s annual event every year since 2022 (which apparently makes me a relative newcomer). And every year there seems to be a new theme. Last year, it was trade and tariffs – a topic that is very much still with us.
That said, what garnered a lot more attention this year – whether during our panel discussion or along the sidelines of the conference – was AI, data centers, and the boost they could provide to steel demand. More on that in a moment.
Let’s first rewind to the Sheet Metal Council meeting last year. There was a lot of discussion about potential tariffs from the incoming Trump administration. Because Trump – even before taking office again – made clear that he would be using them.
Back in December 2024, some people wanted to talk about what those tariffs might look like. Others said they’d rather focus on something concrete. Namely, the sweeping AD/CVD case filed by US mills earlier that year against imports of coated flat-rolled steel.
This year, the debate around tariffs and the coated trade case continued. The consensus seems to be that there will be no TACO (“Trump Always Chickens Out”) when it comes to steel. (Editor’s note: SMU and AMU will explore the TACO/no TACO topic further on Wednesday, Dec. 17, in a Community Chat webinar with Wells Fargo Managing Director Timna Tanners. You can register here.)
So, what happens when you add high duties stemming from the coated AD/CVD case to Section 232 tariffs? (Final margins are here, btw.) A debate about whether galvanized imports will have a significant role to play in the US market or whether new domestic capacity means there will be less room for galv imports in the future. (There is a lot of new galvanizing capacity.)
There didn’t seem to be much concern about the willingness and ability of domestic producers to make heavy-gauge product. The questions centered more around US mills’ commitment to the light-gauged galvanized market. Light-gauge galv is used in a lot in HVAC, especially in duct work. And the market in the past has largely been served by imports.
On the one hand, some folks said domestic mills have assured them it’s different now. With the capacity and capabilities they’ve added (or will add soon), US mills should have the ability to consistently supply light-gauge galvanized buyers. On the other hand, others said they’d received similar assurances from US mills in the past – only to see domestic mills back away from making light-gauge galv when the market improved. Basically, why make light gauge (which requires more time on the mill) when you can make heavy gauge (which requires less time on the mill) and make more money?
Which brings me to AI
Depending on who you talk to, you can come away with the impression that the market might soon see a test of US mills’ commitment to light gauge. Why?
There was a palpable sense of optimism among folks at the HARDI conference about demand from data centers. In other words, around the physical infrastructure – which also includes energy transmission – needed to feed the AI boom. The big question: Could demand from data and AI be far stronger than the market yet realizes?
After all, data centers are steel intensive, maybe especially when it comes to their use of galvanized products. Galv is used in everything from HVAC systems to joists, decking, and rack. The American Galvanizers Association has a good article here on the galv required in data centers. And Nucor has a really helpful graphic here.
You can see why folks who’ve landed work on such projects are excited. A couple said deep-pocketed tech companies driving demand for data centers might care less about the price of steel than securing availability – a refrain I haven’t heard for a while. If you believe those folks, the word “allocation” could come back into the picture next year.
Yes, but…
But, as with all things AI, there are skeptics as well. They question whether data centers truly represent a new demand driver and whether the AI boom is sustainable. Some say the market now feels uncomfortably close to the dot.com boom of the mid/late ’90s – which of course was followed by the dot.com bust in 2000-01.
It’s not that they doubt that the potential for AI to reshape our world. It’s just not yet clear who the winners and losers will be. Besides, some might add, while mills might be insisting on higher prices now, few are telling customers they can’t get the steel they need.
Mind the gap
And in the short term, a bigger topic might be the narrow (by post-pandemic standards) spreads between hot-rolled (HR) coil prices and galvanized base prices.
We saw NLMK USA make an attempt last month to establish a premium of $150 per short ton (st) for galvanized base prices over HR prices. That move appears to have been roughly followed on Wednesday by Nucor. The Charlotte, N.C.- based steelmaker said its new base price for galv would be $1,050/st – or $120/st above its list price of $930/st for HR.
Meanwhile, we continue to see US mills adjust coating extras higher. Zinc prices have been up lately, which is the obvious explanation for higher extras. It might also be a way to increase profits from galvanized sales even if the narrow spreads prove to be stickier than expected.
And that’s another big question. What’s driving those narrow spreads? Is it a structural shift, the result of so many US mills adding new galvanizing capacity? Or is it just a case, as we’ve seen in the past, of galvanized prices following HR higher or lower on a lag?
Will the envelope be pushed too far?
Finally, it’s become consensus that 50% tariffs combined with high AD/CVD duties should limit the participation of imports in the US – both in general and when it comes to galv. But at some point, as US mills test how high they can raise prices, do they push prices a little too far and invite imports back in?
It didn’t seem like many people expected that to happen soon. Why? Well, partly because US trade policy is so unpredictable. President Trump – or any US president for that matter – could raise or lower tariffs with a social media post, making the risk of importing greater than it was in the past. But it’s one to keep an eye on.

