Wolfspeed Inc. WOLF

Key data

The business

Wolfspeed makes silicon carbide, from the raw crystal boule through the polished wafer to the finished power chip. Silicon carbide handles higher voltage and heat than ordinary silicon, so it goes into the parts that move big power: electric-vehicle inverters and onboard chargers, fast-charging stations, grid and factory equipment, and, the newest line in the 10-K, the power supplies feeding AI data centers. It is one of the few vertically integrated players and pushed first to 200mm wafers, which cut cost per chip once volume shows up. That last clause is the whole company. Wolfspeed borrowed about $6.5 billion to build two enormous 200mm plants, the Mohawk Valley device fab in New York and the substrate plant in Siler City, North Carolina, on the belief EV demand would fill them. It did not arrive fast enough. The debt crushed the company, and in September 2025 it emerged from Chapter 11 with roughly $4.9 billion of that debt swapped into equity, wiping out the old shareholders. The moat, if there is one, is the substrate: few rivals can grow 200mm silicon-carbide crystals at scale.

The numbers

Revenue is going the wrong way for a company trying to fill a fab.

Sales fell in every quarter of fiscal 2026, and the just-reported fourth quarter (ended June 28, filed August 20) held near $150 million with a $2.26 loss per share. Gross profit stayed negative: Wolfspeed loses money at the factory gate before a dollar of overhead. Gross margin ran near negative 39% in September, negative 46% in December, then negative 27% and about negative 25% in the two latest quarters, and negative 35% on a trailing-year basis. The margin improved while revenue fell, which means the better loss came from cost cuts, not a fuller fab. The utilization thesis needs volume up. It is down.

Across the fiscal years the arc is a business that grew revenue to an $807 million peak in FY24, slipped to $758 million in FY25, and landed near $665 million in FY26 while gross margin flipped from plus 10% to deeply negative. The emergence cut long-term debt by about 74%, the single real accomplishment of the year. This is not a compounder to model, it is a survival-and-inflection binary. What the market prices is a busted EV silicon-carbide story. What it is not paying for is whether AI data-center power fills the fab on demand nobody underwrote at bankruptcy. The print that settles it is sequential revenue turning up while gross margin crosses toward positive. Cash fell from $1.03 billion after the emergence financing to $576 million at year-end, about $226 million a quarter, against $1.69 billion of debt. Capital spending was cut hard, from roughly $104 million a quarter to $30 million, which slows the bleed, but the math is plain: without a revenue inflection, the cash that bought survival is a clock, not a moat.

Management

New CEO Robert Feurle took over in 2025, and the board and capital structure were rebuilt at emergence, so today's owners are the creditors who took equity, not the shareholders who financed the buildout. The only Form 4 activity in the last year is Citigroup, a holder from the restructuring, buying about $4.5 million and selling about $5.1 million within two days in late October, position management with plan status not disclosed. No insider is stepping up with personal money. Pay for the new team is not in a filed proxy yet, so the record to judge is the last team's, which bet the company on a demand curve that did not arrive, on borrowed money, and lost it.

How it fails or surprises you

Liquidity runs the clock down. Cash dropped from $1.03 billion to $576 million in two quarters, about $226 million each, against $1.69 billion of debt. Capex was slashed to $30 million a quarter to slow it, but if revenue does not inflect within roughly a year, Wolfspeed faces another raise, dilutive at this price, or a second restructuring. Watch cash and free cash flow the next two quarters.

The margin turn is not from volume. Revenue fell every quarter of FY26 while gross margin improved off cost cuts, not a fuller fab. At about negative 25% today with sales still sliding, the line to positive looks steep, and a miss confirms the fab is structurally underloaded.

AI data-center power (right tail). The 10-K adds AI data centers to the demand list for the first time. High-voltage DC power in the rack is moving toward silicon carbide, a market nobody underwrote when Wolfspeed went bankrupt on an EV thesis. If that content ramps faster than EV weakness bites, the empty New York fab fills on demand the stock is not paying for. Watch industrial and energy revenue and any named data-center design win.

Closing thoughts

This is not priced on earnings, there are none, it is priced on whether the equity survives to see the fab fill. What survives even in the bad case is the substrate franchise and the 200mm capacity, which is why the enterprise still carries about $2.6 billion while the stock sits at $28. The fatter tail is the downside: revenue is falling, gross margin is deeply negative, and cash is a clock. If liquidity breaks, most of the equity is at risk through dilution. If AI power and EV both turn while the fab is nearly free, the upside is a multiple of today's price, because a full 200mm fab at positive margin on this revenue base is a different company.

The bet is still that electric cars and AI data centers need more silicon carbide, and that Wolfspeed's giant plants fill up enough to earn money on each chip instead of losing it. It breaks if revenue keeps falling. The one pair to watch is sequential revenue and gross margin, and until revenue turns up and gross margin crosses toward negative 10%, the fab is not filling and the equity is the wrong instrument for the story.