On Wednesday, Federal Reserve chair Kevin Warsh might effectively take the other side of the bet.

It’s been a hot American summer. Oil is hot, hovering around $110 a barrel. Bond yields are hot, too: the 10-year Treasury yield has pushed above 5%, around its highest level since 2007. Credit markets are running hot as well: U.S.-dollar debt issuance to finance AI and data-center development reached $308 billion through July. And all that borrowing is competing with U.S. national debt, which crossed $40 trillion less than a month ago. Stocks, despite a rough few days, are still up roughly 11% this year. Inflation, meanwhile, remains above 3%.

Put all that heat together, and the Federal Reserve is staring down a question it hasn’t seriously confronted in three years: Is the U.S. economy actually overheating? Markets are betting the Fed thinks the answer is at least “maybe.” Traders have priced a quarter-point hike Wednesday with near certainty.

But whether Wednesday amounts to a one-time course correction or the beginning of a new tightening cycle depends on what, exactly, is making the American economy hot. The last time the Fed began raising rates, in March 2022, Jerome Powell’s Fed ultimately raised its benchmark rate by 525 basis points over 16 months.

Mohamed El-Erian, Wharton professor of practice and chief economic adviser at Allianz, parsed the current fervor and anxiety into four questions on X Tuesday: whether oil-supply disruptions persist, with China potentially acting as a “swing consumer”; whether Treasury Secretary Scott Bessent intervenes again to influence long-end yields; whether this week’s hike proves “one and done” or the beginning of a cycle; and how markets balance AI’s enormous promise against its enormous risks.

The ultimate question is whether the inflationary period we’re experiencing is due to an unusual pileup of supply shocks, or evidence that aggregate demand is running too fast for the economy to handle.

Jon Hilsenrath, the former Wall Street Journal Fed reporter and founder of Serpa Pinto Advisory, falls in the overheating camp. His evidence comes from his preferred metric: nominal GDP, the total dollar value of what the economy produces, without adjusting for inflation. Nominal GDP grew about 6% from a year earlier in the first quarter and more than 6.5% in the second, he told Fortune.

If the economy produces about 2% more goods and services every year, and the Fed wants prices to rise about 2%, then nominal growth around 4% would be equilibrium. At 6% or 7%, something has to give: either America has unlocked an unusual productivity boom, or there’s too much demand for the amount of stuff being produced.

“It sure does look like the economy is overheating,” Hilsenrath said. He pointed to several proximate causes: a federal budget deficit running around 6% of GDP, the historic AI investment boom, and the delayed effects of 175 basis points of rate cuts in 2024 and 2025, all hitting at once.

But Goldman Sachs sees almost the opposite economy. Its economists argue there’s “not a strong economic case” for hiking at all. Their “Bottlenecks Tracker” looks for factory-capacity constraints, labor shortages and wage pressures—the usual symptoms of overheating. But those constraints are now slightly less widespread than before the pandemic, save for a couple of industries closely tied to the AI boom.

Goldman argues much of today’s inflation overshoot comes from tariffs and other supply shocks that higher interest rates don’t fix. “The economy is not overheated,” its economists wrote, “which is the usual rationale for raising rates.”

This is why El-Erian pointed to all four factors as question marks. Oil could be a temporary supply shock that fades without help from the Fed, or a persistent disruption that works its way into inflation expectations. The bond market could already be doing the Fed’s work, since a 5% 10-year pushes up borrowing costs; or the yield spike could be a warning that inflation, deficits and debt issuance are becoming too entrenched.

And then there is AI. If the hundreds of billions poured into data centers lift productivity, then the old 2% “speed limit” could be too low. But in the short run, the same boom is an enormous investment-demand shock, increasingly financed through credit markets and concentrated among companies rich enough that another quarter-point hike may barely change their plans.

Looming over the decision is Bessent’s Treasury, which has expanded buybacks of longer-dated debt, despite the fact it knows the Fed is considering tightening at the short end. That leaves the two institutions playing a game of tug-of-war ondifferent parts of the same yield curve.