Inspector's Eye Reads Past AI's Sticker Price

A good home inspector never trusts a fresh coat of paint. He walks through the kitchen, notes the granite countertops, and then heads straight for the crawl space. The reality is in the joists and wiring—the bones the stagers miss. A house can look like a million bucks and still be rotting underneath. And a house with peeling wallpaper can have a foundation built to outlast the next three owners.

Wall Street, according to a recent analysis from the investment research firm Being Exponential, behaves a lot like a nonchalant homebuyer. Investors see a stock trading at 100 times earnings and they run. They see a chip name with a price-to-earnings ratio that looks like there’s nothing left for them. Earnings, on their own, are just the countertops. They get dressed up by depreciation schedules, debt loads, and amortization assumptions that have little to do with how the actual business is performing.

To see the foundation, you climb underneath. You look at EBIT. You look at EBITDA.

Why the optics trade looks cheaper than the sticker price suggests

EBIT is earnings before interest and taxes. EBITDA adds depreciation and amortization back into that number. The report looks at EBITDA for nearly every company it covers, because it strips out accounting noise and shows the cash the business is actually throwing off. Run that lens over the optics trade, and the picture does shift.

Credo Technology Group (CRDO) trades at 39 times forward earnings, 34 times forward EBIT, and 33 times forward EBITDA. On a standalone basis, that sounds rich. Pair it with 82% revenue growth this year, 47% the year after, and earnings-per-share growth running as high as 130%, and suddenly 30-something times EBITDA looks like a bargain.

Astera Labs (ALAB) is the more expensive cousin, at 103 times forward earnings and roughly 100 times forward EBIT and EBITDA. The growth underneath is enormous: 81% revenue growth this year, with EPS growth north of 140% layered on top of stable margins.

According to the analysis, it doesn’t matter whether you’re looking at price-to-earnings, EBIT, EBITDA, or revenue-to-book. The optics trade is cheap relative to the growth profile sitting on top of it. The report calls that convergence—cheap multiple, massive grower—exactly the setup it wants to own.

A critic might note that EBITDA ignores capital expenditures, which can be significant for hardware companies scaling production. The analysis itself does not address that caveat, but it’s worth keeping in mind.

Redwire: the dilution dip

Redwire (RDW) issued stock after a hot run, and the stock pulled back. Stock issuances create short-term pain and long-term opportunity, the report argues, because good companies use that capital to build more stuff and generate more returns. Technically, Redwire is sitting on a major shelf: the January high around $14.20, the 50-day moving average at $13.58, and multiple historical lows clustered in the $13.80 to $13.90 range going back to 2025. The firm likes the $13 to $15 zone to buy the dip.

BWX Technologies: the nuclear comeback

BWX Technologies (BWXT) is the industrial backbone of U.S. nuclear power, supplying the Navy’s propulsion program for decades and now riding the AI data center power crunch. The bear case is orbital compute—solar-powered satellites that don’t need nuclear at all. The report’s counter: that future is probably a decade away from going mainstream, and it will never fully obsolete the data centers already on the ground. The structural power problem driven by terrestrial compute persists for the next five to 10 years, minimum, and likely longer. Technically, BWXT lost its 200-day moving average, bottomed in the 180s, and has U-turned back above it on a near-oversold bounce—a pattern that looks a lot like early-2025’s setup before a major rally.

Bloom Energy: the backup power king

Bloom Energy (BE) is becoming a go-to provider of on-site backup power for AI data centers. The growth numbers back it up: 83% revenue growth this year, 73% next year, with EBITDA margins expanding from 20% toward 30% by the end of the decade. Valuation is rich—92 times forward earnings, 72 times forward EBITDA—but the trailing 12-month average EBITDA multiple is 61.5 times, with one standard deviation above sitting near 83 times. The stock just bounced off that 60-times level in a V-shaped recovery. The report says it is buying dips toward 60 times and fading rips toward 100 times.

Micron Technology: five steps forward, three steps back

Micron Technology (MU) has run hard since summer 2025, with 20% pullbacks showing up in August, November, March, and again in June. Each one was a buying opportunity, according to the analysis. MU stock just hit new highs near $1,080 and pulled back slightly. The rule of thumb offered: accumulate on 20% pullbacks, because there will be more of them.

The verdict across all five names is the same. Don’t judge the house by the paint job. Climb into the crawl space, check the growth underneath the multiple, and buy the dip when the foundation is sound.