If you are waiting for mortgage rates to fall, there is a reason the wait keeps getting longer, and it has almost nothing to do with housing. The Federal Reserve went from talking about rate cuts to talking about rate hikes, and when you read what the hawks on the committee actually said, one thing comes up over and over: AI spending.
The Fed flipped, and that is the whole story
Earlier in the cycle, the market was pricing in two to three rate cuts. That expectation has now reversed to two to three potential hikes. That is not a small adjustment. It is a complete change in direction, and it happened without housing data getting better or worse in any dramatic way.
So what changed? Read the statements from the hawkish members of the committee and a pattern emerges. Names like Beth Hammack and Neel Kashkari keep returning to the same theme, and Lorie Logan has addressed it too. Their argument is that AI investment is inflationary.
Why AI spending is treated as inflationary
A lot of people assumed AI would go the other direction. The popular fear was that AI would take everyone's jobs, unemployment would spike, and fewer people would be able to buy homes. That is not what the data has shown, and it is not how the Fed is reading it.
Here is the mechanism they are describing:
What about the productivity miracle?
The counterargument is that AI will eventually deliver a productivity boom, and productivity growth is disinflationary. Several Fed officials have said as much.
But "eventually" is doing a lot of work in that sentence. The old-school model is straightforward: if you want to hit 2% inflation, you need wage growth under 3%, because productivity runs around 1%. Productivity has picked up. It has not picked up in any miraculous form. Betting your home purchase on a productivity miracle that has not arrived is a very expensive way to be right eventually.
Here is why this is a housing story
This is the connection most people miss. Roughly 65% to 75% of where the 10-year yield and mortgage rates go is Fed policy. Not housing supply. Not buyer demand. Fed policy.
So if the hawks on the committee are naming AI spending as the reason for their stance, then AI spending is a mortgage rate issue by direct transmission. You do not need a theory about data centers lowering nearby home values to make AI a housing problem. The Fed's reaction function alone does it.
It reaches housing through three separate channels:
The uncomfortable part
The Fed's position is that policy is not restrictive for the general economy, because AI is supporting growth and investment.
It is absolutely restrictive for the housing market. Commercial construction has stalled, single family starts and permits are down, and apartment development is flat. Housing is the sector feeling the squeeze while the broader economy looks fine on paper.
The Fed knows this. Housing is simply not the variable driving the decision.
So where does that leave rates?
Let's be specific rather than vague, because vague is what leads people to wait indefinitely.
If Fed policy settles around a 3% neutral rate, it becomes very difficult for mortgage rates to break below 5.75%. And the range we have actually lived in this year has been roughly 6% to 6.85%. That is the world we are in.
Why this outlasts the conflict
This is the part that should reframe how you think about waiting. The conflict has a potential end date. Geopolitical resolution, political pressure, or simple exhaustion could finish it, and rates could move quickly when it does.
AI investment has no end date on the horizon. The global race to build is accelerating, not slowing. Countries with aging populations are betting on AI and robotics to replace labor they no longer have, which means the spending has a structural motive behind it rather than a cyclical one.
Structural, not temporary
Even if AI is the smaller variable compared to the conflict, it is the more durable one. The conflict could end tomorrow and rates could fall. AI investment will not end tomorrow.
That distinction matters more than the size of either factor, because it changes what you are actually waiting for.
What could actually change it
There is a real path lower, it just requires things that have not happened yet:
- The conflict resolving. Many expected an end well before now. It has not come, and it is arguably a bigger variable than AI.
- Genuine productivity gains showing up in the data, not in projections. That is what would make AI disinflationary instead of inflationary.
- A meaningful slowdown in AI investment. There is no sign of this. By most accounts we are at the beginning of that spending cycle, not the end.
- Labor market softening enough to change the Fed's tone. Data center construction is currently working against that.
The political wildcard
There is a genuine backlash building, and it is bipartisan because electricity prices hit everyone. Governors in both red and blue states are placing moratoriums on new data centers, and one poll put negative sentiment toward AI around 75% of the population. Local officials are absorbing constituent anger over power bills, noise, land use, and a lack of transparency in how these deals got made.
The AI industry also damaged itself: leading with "this will take your jobs" before showing any benefit was a poor opening argument.
History is fairly consistent here: politicians do not survive inflation. When something large injects money into the economy and prices rise, the public assigns blame and politicians respond by regulating or restricting it. Midterms are the natural pressure valve, and this is a 2026 story, not a 2027 one.
Why a real moratorium would matter for housing
If the backlash gained enough momentum to meaningfully slow data center construction, it would do two things at once: remove one of the Fed's stated justifications for staying aggressive, and potentially redirect construction labor back toward homes and apartments.
That would be genuinely significant for housing. The honest read is that we are not there yet.
There is one more irony worth naming. The stated economic goals have been a lower dollar, lower energy prices, and lower mortgage rates. AI investment currently works against all three at once: it holds rates up, pushes electricity prices up, and supports the dollar through investment inflows.
The bottom line
Mortgage rates are not dropping meaningfully anytime soon, and the reason is not housing. The Fed reversed from expected cuts to potential hikes, its hawkish members keep naming AI spending as inflationary, and Fed policy drives 65% to 75% of where mortgage rates go. That makes AI a housing problem whether or not a data center is ever built near you.
And unlike the conflict, this variable has no visible end date, which makes it structural rather than temporary. Practically, that means planning around a 6% to 6.85% range with roughly 5.75% as a realistic floor, rather than waiting for a number that current policy does not support. If the math works on the home in front of you today, that is a real decision. You can refinance a rate later. You cannot renegotiate a purchase price you never made.
Should you buy at today's rates?
Let's run your actual numbers against today's market instead of a rate you are hoping for, and see whether the payment genuinely works for you.