BRUN, DGXX, & AI Infra Comps Analysis

🌾 Welcome to StableBread’s Newsletter!

It's been about a month since StableBread transitioned into a research-focused newsletter, so I wanted to provide a quick portfolio update.

Near the end of every post I perform a deeper financial, valuation, and often modeling exercise, then share whether I actually bought shares in the stock.

Of the 6 write-ups released over the last ~30 days, only two ended with a buy recommendation: (1) Boost Run (WLAC/WLACW; now BRUN/BRUNW) and (2) DigiPowerX (DGXX).

Boost Run was proposed on April 12, when commons (WLAC/BRUN) were at $10.70/share, and warrants (WLACW/BRUNW) were just under $3.00/share.

If you invested $1,000 and deployed half ($500) into WLAC/WLACW then (equally), you’d be up ~$442 (+177%) and ~$1,226 (+490%) respectively.

DigiPowerX was proposed on April 29, when the stock was trading at $3.50/share.

If you deployed the other half ($500) into DGXX then, you’d be up ~$577 (+115%).

Therefore, investing just $1,000 would’ve returned ~$2,245 (+225%) in gains (total current value: $3,245).

Obviously, these returns are a bit abnormal given the current market environment and returns aren't guaranteed, but the goal is to find a few asymmetric setups a year where the work justifies a sized position, and let those compound over time.

The rest of this post covers three topics:

Boost Run started trading on Nasdaq as BRUN on May 11, 2026, replacing WLAC after the April 30 merger vote.

The common sits at ~$30/share, up ~184% from the trust value, and the warrants (BRUNW) are at ~$17.70, well above the $14.50 breakeven.

What you should know is that most of the company isn't actually tradeable yet.

BRUN has 61.43M shares outstanding at close, but only 12.65M will trade for the first 6 months. The rest sits behind a 6-month lockup from the May 11 close, expiring in mid-November 2026.

Zero redemptions on April 28 meant all 12.65M public WLAC shares rolled into BRUN. That's the entire float right now.

The Sponsor's 4,628,674 founder shares and the Boost Run sellers' 44.15M sit behind the 6-month lockup.

The lockup has one exception: 10% of each group's shares release early if BRUN closes at or above $12.00 for any 20 trading days within a 30-day window after close. That's 462,867 from the Sponsor and 4.42M from the Boost Run sellers, or ~4.88M shares total.

At $30, the $12 condition clears easily. But the observation window only started ticking on May 11, since WLAC's pre-close price action doesn't count. So expect those early releases 30-45 trading days out.

That's meaningful against a 12.65M float, but not catastrophic. The bigger supply event is the full 6-month lockup expiration in November.

CEO Andrew Karos can also receive up to 7,875,000 earnout shares across three equal tranches of 2,625,000. Each tranche vests if VWAP holds at or above the target on 20 of any 30 trading days:

VWAP is the volume-weighted average price, basically the average daily price weighted by volume at each level.

The Sponsor and the SPV (a separate vehicle that bought part of the Sponsor's stake) get another 3,093,750 shares on the same tiered structure, putting total contingent earnout issuance at 10,968,750 shares.

At $30, all three tiers sit well above their hurdles. If the price holds anywhere near here, all three should trigger within the first 60-90 trading days.

Beyond the earnouts, there are 6,325,000 public warrants outstanding under BRUNW. Each gives the holder the right to buy 1 BRUN share at $11.50, with five years until expiration.

There are two paths from here.

The first is a cash exercise. The warrant holder pays $11.50 and the company issues 1 new share. If all 6.325M warrants exercised this way, BRUN pockets ~$73M in cash and issues 6.325M new shares.

The more likely path is forced cashless exercise. This is less dilutive than a cash exercise, and BRUN probably doesn't need the extra $73M given the $132.6M trust proceeds already received at close.

When BRUN closes at or above $18.00 for 20 of any 30 trading days (after a 30-day waiting period from close), the company can issue a redemption notice at $0.01/warrant. Holders are then forced to either exercise or effectively lose the value.

Under cashless mechanics at $30, each warrant converts to ($30 − $11.50) / $30 = 0.617 shares. That's 6.325M × 0.617 = ~3.9M new shares, with zero cash to the company.

The $18 condition is already cleared on price. The company will almost certainly call redemption once the 30-day post-close waiting period passes and the 20/30 observation window completes.

One detail worth flagging: The 5,145,722 private warrants outstanding (4,007,222 held by the Sponsor and 1,138,500 by the IPO underwriters, BTIG and Craig-Hallum) aren't subject to the $18 forced redemption.

They sit as long-dated optionality, exercisable anytime over the next 5 years. That's an overhang that doesn't trigger on any mechanical condition.

By the time the 6-month lockup expires in mid-November 2026, the share count will have changed. Starting at 61.43M and layering in the contingent issuances:

That puts total shares outstanding in the 76.3M to 78.7M range. At lockup expiration, 43.9M previously locked insider shares unlock at once.

The tradeable float jumps from ~32M (after the early releases, earnouts, and warrant exercises trigger) to ~76M+ essentially overnight.

Even a modest 10% sell-through from insiders is ~4.4M shares hitting a ~32M float. That's ~14% of the available float coming up for sale.

For the first 6 months of BRUN trading, only ~16-41% of total shares outstanding are actually tradeable.

The low end (16%) reflects the initial 12.65M float against the ~78.7M ultimate share count. The high end (~41%) reflects the float after the early releases, earnouts, and warrant exercises all trigger, but before the main 6-month lockup expires.

A small float combined with an AI infrastructure narrative and stacked mechanical supply events tends to introduce volatility in both directions. Expect sharp moves on relatively modest news flow over the next 6 months.

Run-up-then-sell-off behavior is common around SPAC lockup releases as insiders distribute newly unlocked stock. Both the early release this summer and the full unlock in November could play out this way, with November being the more material event.

None of this changes the fundamentals of the business or the underlying thesis. But understanding the share structure helps contextualize the price action you'll see in the coming months.

What this means if you bought warrants (WLACW/BRUNW)

Each warrant has $18.50 of intrinsic value at the current stock price ($30 − $11.50). Under forced cashless exercise, BRUN issues you shares worth $18.50 at the $30 share price, which works out to 0.617 shares per warrant.

Once BRUN issues a redemption notice (likely within the next 60-90 trading days), there's a 30-day notice window. If you do nothing during that window, your BRUNW automatically converts to 0.617 BRUN shares per warrant at the end of it. No new cash required.

At $17.70 today, the warrant is already trading essentially at its cashless conversion value (0.617 × $30 = $18.51).

The asymmetric leverage that made WLACW attractive at $3 (when the breakeven was $14.50) is gone. From here, BRUNW moves dollar-for-dollar with BRUN (the 0.617 conversion ratio applies if redemption is called at today's price; it scales up or down with BRUN before then).

That leaves two real choices:

Selling now and rebuying common with the proceeds is strictly worse, since $17.70 / $30 = only 0.59 shares per warrant. Selling outright also triggers a capital gains tax event, while holding through cashless conversion is typically non-taxable (considered “recapitalization”).

DigiPowerX (DGXX) dropped Q1 2026 results on May 15, 2026. Here’s what management shared (my analysis follows):

Financials:

Balance sheet:

AI / Operations:

Strategic:

Forward run-rate guidance from CEO:

Cerebras (CBRS) IPO’d on Thursday, opening at $350/share, up from its $185 IPO price. CBRS alone supports the thesis for our early DGXX investment.

But share count moved from 69.8M to 90.4M between March 31 and May 15. Most of that came from the $103M ATM raise at ~$5.16/share avg.

The cash gives Phase 1 self-funding optionality, but it's worth recalibrating per-share assumptions in any model that didn't anticipate this much dilution this fast (including ours).

Beyond that, the May 8 prospectus (ATM equity program) authorized up to $175M in ATM capacity. They've used $103M, leaving ~$72M still available to tap. More dilution optionality is sitting there.

Phase 2 financing is the next thing to watch. The 10-Q states the additional 25MW is "conditioned on the Corporation securing adequate financing."

The 70/30 debt plan is the path to making it happen. If it works, dilution slows considerably. If it doesn't and they tap the rest of the ATM, that's another ~12M shares on top of the 90.4M already out.

Management's 2027 $300M run rate target assumes Cerebras hits the full 40MW on time, plus another ~50MW of non-Cerebras colo customers, plus GaaS scaling from today's 1MW (SubQ) to 10-12MW.

Execution risk is real, but nothing suggests they're behind schedule.

For now, I'm watching how the debt financing terms come together over the next couple quarters.

I ran a comps valuation on 10 publicly traded AI infra operators using close-of-day data from May 14, 2026 (one day before DGXX's Q1 earnings).

Names included: NBIS, CRWV, IREN, DGXX, CIFR, WULF, BRUN, APLD, BTDR, WYFI.

The comps set isn't perfect (mixes neoclouds w/Bitcoin miners pivoting into HPC and AI hosting), but it provides a useful market-based benchmark.

Financials are calendarized to CY2026E so the comps are more apples-to-apples.

Here's what I found:

Starting with market data, EV ranges from $410.7M DGXX (even after the CBRS $1.1B deal + IPO) to $95.2B CRWV.

CRWV is the only highly levered name in the group, carrying $32.9B in net debt which is ~35% of its EV.

Everyone else runs cleaner balance sheets, and DGXX and WYFI sit on net cash.

Most names trade near 52-week highs too. Only CRWV (61.1%) and BTDR (53.1%) are well off.

On financial metrics, the whole set/narrative is small revenue today --> huge revenue projected.

Calendarized 2026E revenue is ~1-7x 2025A across the group. The largest jumps are NBIS and BRUN.

Adj. EBITDA flips from negative or near-zero in 2024A/2025A to profitability in 2026E and 2027E (across all comps).

Take Adj. EBITDA with a grain of salt. Every company defines its own add-backs (SBC, impairments, crypto fair value swings, restructuring).

Numbers are pulled from Fiscal.ai (except BRUN which is from the S-4). Reference the actual 10-Ks if you want tighter numbers.

On revenue growth, from 2025A to CY2026E, the median grows 123.7% and the mean grows 215.4%.

From CY2026E to 2027E, the median grows 131.2% and the mean grows 144.6%.

BRUN leads 2026E at 569.4% off a tiny base. NBIS closely follows at 540.5% with a CY2026E revenue base almost 19x larger!

CIFR is the slowest at 5.8% before re-accelerating to 273.6% in 2027E as new HPC capacity comes online.

Margins are where the thesis eventually lives.

The Adj. EBITDA margin in CY2026E comes in at 40.9% median and 41.5% mean. In 2027E that expands to 62.5% median and 57.3% mean.

BRUN leads at 66.0% in 2026E (S-4 projection) and 72.4% in 2027E (Fiscal.ai projection). CRWV, IREN, WULF, and WYFI all cluster in the low-60s by 2027E.

Margins are driven by where each name is in their HPC buildout curve and the cost structure underneath it (owned vs colocated capacity, power contracts, hosting vs self-operated mix).

On trading multiples, EV/Revenue is the cleaner read here.

Every company defines its own add-backs on Adj. EBITDA, and several names are still ramping out of negative or near-zero territory in 2026E.

EV/Revenue median sits at 10.8x in CY2026E and compresses to 4.7x by 2027E.

Mean is 18.8x and 7.9x respectively. The compression from 2026E to 2027E tells you the market is pricing in the revenue ramp pretty aggressively.

Notice the spread w/in CY2026E as well. NBIS trades at 16.6x while DGXX sits at 7.5x.
Growth is the main differentiator. NBIS is the highest grower in the set at 540.5%, DGXX is mid-pack at 59.3%.

Look at DGXX on 2027E though, it drops to 2.0x. Cheapest on that basis.

EV/Adj. EBITDA median is 22.5x in 2026E and 8.6x in 2027E. Means of 70.9x and 14.2x are skewed hard by CIFR (379x) and WULF (116x) where 2026E EBITDA is depressed.

Median is the only signal worth using here.

Now to implied valuation.

I own three names in the set: NBIS, DGXX, and BRUN.

Let's apply implied median 2026E multiples to each company's projected CY2026E metrics...

NBIS at $221.15:

Trades at a meaningful premium on both. Justified by NBIS posting the highest 2026E revenue growth in the set at 540.5%. And I'd argue the highest-quality business.

DGXX at $7.22:

Mixed read. ~38% upside on EV/Revenue, ~10% downside on EV/Adj. EBITDA. Trades like a name where the market is pricing in revenue scaling but discounting margin durability.

BRUN at $26.04 (using ~72.2M post-merger shares per the S-4):

Fair on EV/Revenue. ~42% upside on EV/Adj. EBITDA.

Hope this helped provide a high-level read on how the market is pricing AI infra stocks.

You can download the model below: