Neocloud Brief

CRWV

CoreWeave

AI compute · GPU neocloud

$81.08

+1.5%

$60.55 low$153.20 high

01 · What do they do?

Rents high-end GPU capacity to AI companies through a debt-heavy cloud platform.

02 Do they make money?

$2.575BQ2 revenue · +112.5% YoY

−25.41%net margin

−$2.68TTM EPS

$12.4–13.2B2026 outlook

Q2 net loss is approximately $654M when the supplied net margin is applied to revenue. Backlog reached $104.2B, up 264% YoY.

Loss-making

03 Can they survive?

$5.5Bcash

~$8.1Bshort-term debt

5.53×debt / equity

$35–39B2026 capex plan

Capex vs. funding: 2026 capex guided at $35–39B versus a ~$12.4–13.2B revenue outlook—spending roughly 3× sales.

Demand is deep, but the buildout requires very large, continuing access to capital.

High leverage

04 What does Wall Street think?

Moderate Buyconsensus

~$140–142average target

~+74%implied upside to midpoint

$41–250low / high target

The enormous target range shows how divided analysts are on execution and financing risk.

05 What’s next?

Q3
2026

Q3 earnings and Nasdaq-100 inclusion. Revenue guide is $3.45–3.60B; investors will watch delivery against backlog and funding needs.

IREN

IREN Ltd

AI compute · miner-to-neocloud pivot

$45.99

+5.8%

$28.93 low$76.87 high

01 · What do they do?

Converts former Bitcoin-mining power capacity into AI cloud infrastructure, anchored by Microsoft.

02 Do they make money?

$70.5MFQ4 AI cloud revenue

+110%sequential AI cloud growth

−$2.22EPS

$4BFY26 ARR outlook · contracted

The AI business is scaling quickly, but the supplied snapshot does not include net income or margin; EPS remains negative.

Loss-making

03 Can they survive?

$14Bfunding in place

$25–30BFY27 capex plan

~$8Bstated residual gap*

$3.6BGPU financing

Capex vs. funding: FY2027 capex is $25–30B, with ~$14B in funding in place and a stated ~$8B gap still to close.

Cash and total debt were not supplied. *The stated gap depends on additional funding not itemized in this snapshot; $14B alone implies an $11–16B gap.

Funding gap

04 What does Wall Street think?

$65JPM · Overweight

$90H.C. Wainwright · Buy

$105Compass Point

$83B. Riley

No reliable consensus average was supplied. These individual targets imply roughly 41% to 128% upside from $45.99.

05 What’s next?

2027
CAPACITY

Late-stage capacity contract talks. A new multiyear frontier-lab contract supports the pivot; Bitcoin mining is expected to wind down by decade-end.

NBIS

Nebius Group

AI compute · full-stack cloud

$222.13

+1.9%

$73.52 low$299.86 high

01 · What do they do?

Runs an AI cloud platform spun out of Yandex, with major Microsoft and Meta GPU contracts.

02 Do they make money?

+454%Q2 revenue growth

40.5%EBITDA margin

$0.33TTM GAAP EPS

$7–9Byear-end run-rate target

Revenue dollars and net income were not supplied. Reported P/E is ~1,366×; that differs from a simple price / $0.33 TTM EPS calculation, so its EPS basis may differ.

Early profitability

03 Can they survive?

$20–25Bcapex plan

$17.4BMicrosoft GPU deal

$3BMeta deal

$57.9Bmarket cap

Capex vs. funding: Planned capex is $20–25B versus a $7–9B end-of-2026 annualized run-rate revenue target.

Cash and total debt were not supplied. Large customer commitments support the build, but the capital program leaves little room for misses.

Capex intensive

04 What does Wall Street think?

Moderate Buyconsensus

$260.20average target

+17.1%implied upside

$340Baird · Street high

Piper Sandler is Neutral at $224. Michael Burry disclosed a short at roughly $212.

05 What’s next?

NOV 10
2026

Next earnings. The key check is progress toward the $7–9B annualized revenue run-rate and delivery on the Microsoft and Meta commitments.

DRAM

Roundhill Memory ETF

Memory · diversified fund

$59.11

+2.3%

$26.14 low$81.34 high

01 · What do they do?

Owns a focused basket of memory-chip makers supplying the AI HBM and storage cycle.

02 Do they make money?

~+156–161%since Apr. 2 launch

17holdings

$26.1Bassets under management

0.65%expense ratio

Fund profitability does not apply. Returns come from holdings including Micron, Samsung, SK Hynix and Sandisk.

Profitable basket

03 Can they survive?

$26.1Basset base

17company exposures

0.65%annual fund fee

ETFno corporate debt thesis

Capex vs. funding: Not applicable—DRAM is an ETF, not an operating company, so it has no corporate capex program.

The relevant risks are memory-cycle reversal, concentration and fund liquidity—not one operator’s balance sheet.

Different risk model

04 What does Wall Street think?

Not applicableno company consensus

—analyst price target

Pure-playfirst memory-focused ETF

#1fastest-growing ETF launch

For an ETF, holding-level estimates, memory pricing and flows matter more than a single consensus target.

05 What’s next?

HBM
CYCLE

AI memory demand is the catalyst. Watch HBM supply, memory pricing and capex from Micron, Samsung and SK Hynix for evidence the supercycle can continue.

At a glance

Different wrappers, different failure modes.

Ticker	Scale	Profit status	Capital pressure	Street view	Next watch
CRWV	$43.7B market cap	−25.41% net margin	Very high · 5.53× D/E	~+74% to avg midpoint	Q3 results + NDX entry
IREN	$16.4B market cap	−$2.22 EPS	High · ~$8B stated gap	$65–105 cited targets	2027 capacity contracts
NBIS	$57.9B market cap	$0.33 TTM EPS	High · $20–25B capex	+17.1% to consensus	Nov. 10 earnings
DRAM	$26.1B AUM	ETF · N/A	Fund-level · not debt-led	No single target	HBM pricing cycle

The fast read

CRWV offers the largest consensus upside but carries the most explicit leverage. IREN’s AI pivot is accelerating, with a material funding gap. NBIS is closest to GAAP profitability but priced for near-perfect execution. DRAM spreads company risk across the memory supply chain, while concentrating exposure to a cyclical theme.