Skip to content

Dossier · NBIS · Watchlist

NBIS · Nebius Group N.V.

LOW Compounder Catalyst · gpu-cloud-neoclouds

Last analysed ·

Current thesis

GPU-cloud ACCELERATING, theses resolved bullish but +100%/6wks parabolic on a short-squeeze, correlated to held HUT pullback-only, not a chase.

Current Thesis

The breakout resolved, ran vertical on short-covering, printed its first distribution day (6/5), held a pullback into the 20-EMA shelf (6/8 +5% off the $210 area), and now carries a fresh non-fundamental leg: Nasdaq-100 inclusion announced 6/12 alongside CoreWeave and Astera Labs. Both bear legs that pinned the name through April closed bullish in May the ARR ramp printed real (5/14 Q1), and the opaque-tenant question died once Meta and Microsoft were named. The GPU-cloud narrative is ACCELERATING and fully cluster-confirmed, and the sell-side dispersion is beginning to narrow (BofA went Buy / $280 on 6/8, the first PT above the tape). What is missing is a new fundamental: every hard catalyst is printed and in-price, and there is nothing on the docket until the ~8/13 Q2 print. The interim drivers are passive flows (index inclusion) and build-out headlines (£1.7B UK), both of which can be one-time bids. After ~+100% in six weeks into a $278 spike, the clean setup is a pullback into the rising 20-EMA that holds and reclaims $223.62 not a chase of the high.

Bull Case

  • 5/14 Q1'26 print: revenue $399M (+684% YoY, +75% QoQ); Nebius AI revenue $390M (+841% YoY), ~98% of group. The "will the revenue actually show up" risk that defined this name is gone.
  • 5/14 ARR/margin inflection: core AI-cloud ARR $1.92B (+54% QoQ); group adj EBITDA $130M (32% margin); Nebius AI adj EBITDA margin 45% vs 24% in Q4'25 operating leverage compounding, not promised.
  • 3/16 Meta $27B/5-yr plus the prior $17B Microsoft contract: two named hyperscaler anchors, first large-scale Vera Rubin deploy targeted early-2027.
  • 6/12 Nasdaq-100 inclusion (with CoreWeave, Astera Labs): forces passive index buying and broadens the institutional holder base; Cramer framed it a "big deal."
  • 6/8 UK expansion: £1.7B across four NVIDIA-powered sites, plus a 10-year multi-megawatt Kao Data deployment agreement backlog and contracted capacity extending past the US hyperscalers.
  • 6/9 AI Robotics / Physical AI Living Lab with NVIDIA's robotics program: gives UK and European robotics startups simulation and synthetic-data access, opening a second demand vertical beyond LLM training.
  • 5/20 Bloom Energy agreement: up to $2.6B service fees for ~250MW guaranteed power; contracted-power guide raised 3GW→4GW. Power is the binding sector constraint and NBIS locked it early.
  • 6/2 Macron/SoftBank-Nebius France (€53B European AI-infra commitment, $61.7B total) positions the company as a sovereign-AI build partner.
  • Smart-money stacking: 5/27 Situational Awareness 5.6% stake, 5/15 Nvidia 13F (1,190,476 sh), 6/1 Jensen Computex praise, 6/8 BofA Buy / $280.
  • Balance sheet: Q1 operating cash flow $2.3B on customer prepayments plus ~$2B post-Yandex cash funds the $20-25B 2026 capex with no dilution overhang the non-consensus edge versus debt-funded CoreWeave and dilution-funded APLD/IREN.

Bear Case

  • Peak public attention: ~+100% in six weeks into a $278 premarket spike (6/2), RSI flagged overbought that session, and NBIS appeared on Benzinga most-searched leaderboards through May.
  • The final leg ran on short-covering, not fundamentals: 6/1 "intense short interest momentum" on Jensen's praise. A squeeze that runs vertical unwinds just as fast when the cover-flow exhausts.
  • Index-inclusion is a known event: the passive bid around the effective date is one-time, and "buy the rumor, sell the inclusion" is a common fade.
  • Sell-side still partly below the tape: BNP $255 Neutral (6/2), DA Davidson $250 Neutral (5/18). BofA's $280 (6/8) narrows the dispersion but does not erase it.
  • Nothing fundamental left to price until the ~8/13 Q2 print interim tape is flow, not earnings.
  • Capex of $20-25B in 2026 is enormous; any slip in power or site build-out (the UK/Kao Data execution) re-opens the bear file.
  • High-beta AI-infra correlation: macro risk-off (6/8 Israel-Iran missile-strike headlines) hits stretched names first.

Setup & Price Structure

  • $278 = 6/2 spike high, the resistance to reclaim on volume.
  • 6/5 = first real distribution day of the move.
  • ~$210 = 5/26 pullback low and the rising 20-EMA shelf; 6/8 shares rose ~5% off a key trading signal here, so the shelf held on its first test.
  • $223.62 = 5/28 key-signal level; reclaim and hold flips momentum back on.
  • $190 = post-print gap-fill, the deeper support that defines the broken-vs-intact line.
  • Structure read: a held 20-EMA pullback plus a fresh passive-flow catalyst is constructive, but entries chasing a re-test of $278 sit in the do-not-chase zone between the gap floor and the spike high. A pullback that holds $210 and reclaims $223.62 is the higher-quality long trigger; a daily close back below $190 is the stand-aside line.

Catalyst Calendar (next 30 days)

  • ~2026-06-22 (est.): Nasdaq-100 inclusion effective / passive rebalance flows around the implementation date.
  • Ongoing June: UK build-out execution headlines (£1.7B, four NVIDIA sites; Kao Data 10-yr deal) and Physical AI Living Lab traction.
  • Watch for sell-side cluster expansion more PT raises following BofA's $280 (6/8) would confirm desks underwriting NBIS as the category winner rather than a momentum trade.
  • No earnings in the window. Q2'26 print ~2026-08-13 (est.) is the next hard fundamental test and sits outside 30 days; an earnings blackout applies within 3 trading days of the confirmed date.

What Would Change Our Mind

  • Bull-break: a daily close below the rising 20-EMA (~$210) followed by loss of the $190 post-print gap-fill marks the momentum leg as broken stand aside, do not average into weakness.
  • Squeeze exhaustion: a collapse in short interest paired with volume drying up signals the cover-driven bid is done; a stretched name with no fresh fundamental mean-reverts hard.
  • Theme rolling to SATURATED: mainstream CNBC saturation plus fresh retail Reddit waves with no new contract/flow catalyst caps upside.
  • Confirmation to upgrade: reclaim and hold above $223.62, then $278 on expanding volume, would re-rate the setup from probe-only to a higher-conviction long.

Correlation Notes

  • Primary peer CoreWeave (CRWV): also added to Nasdaq-100 on 6/12 and soaring into it; debt-funded, with the 5/18 margin PT-cut as the leading bear tell for the neocloud cohort.
  • Higher-beta neocloud proxies APLD and IREN (dilution-funded) amplify both directions of any NBIS move.
  • Bloom Energy (BE, +228% YTD) is the power read-through; weakness there flags AI-power-theme profit-taking spilling into the cloud names.
  • The whole complex tracks NVIDIA sentiment Jensen's 6/1 Computex praise moved the group as one.
  • Single-factor caution: NBIS, CoreWeave, APLD, IREN, BE and the bitcoin-miner-turned-AI-host names (HUT/IREN) are one GPU-compute / AI-power trade wearing different tickers. Stacking NBIS on top of existing neocloud or AI-power exposure concentrates a single macro factor, and the squeeze overlay means correlated drawdowns arrive fast size tighter than the standalone chart suggests.

Notes

  • 2026-04-18: seed: Serenity/attention list
  • Earnings blackout: defer any act within 3 trading days of confirmed Q1'26 print date (~2026-05-14 to 2026-05-22 est.)
  • Pair-trade reference: CRWV primary, APLD/IREN higher-beta proxies
  • Apr 13 ATH is THE pivot reclaim on volume = long trigger; close below Apr 8 Cramer-gap = stand-aside trigger
  • Balance-sheet differentiator (~$2B cash
  • no dilution need) is the non-consensus leg of the bull case vs. debt-funded CRWV
  • Balance-sheet differentiator (~$2B cash
  • no dilution need) is the non-consensus leg vs. debt-funded CRWV
  • Mid-range between Cramer-gap floor and Apr 13 ATH = do-not-chase zone; wait for tape to pick a side
  • Wolfe Peer Perform (2026-04-16) is the bear sell-side anchor watch for cluster expansion as leading tell
  • 2026-04-18 seed: Serenity/attention list
  • Q2'26 print est. ~2026-08-13 next hard fundamental test of the ARR ramp; earnings blackout applies within 3 trading days of confirmed date
  • Conviction LOW is an ENTRY-TIMING call, not a thesis call: narrative is ACCELERATING but +100%/6wks into peak-public-attention = chase-trap. Re-rate to MEDIUM/HIGH on a 20-EMA pullback that holds and reclaims
  • Squeeze overlay (behavior) active since 2026-06-01 Jensen/short-interest surge sizing should stay tight even on a pullback entry; a confirmed short-cover exhaustion unwinds fast
  • Pair/peer map: CRWV primary (debt-funded, 5/18 margin PT-cut = leading bear tell), APLD/IREN higher-beta proxies, BE two-way power read-through
  • Key levels: $210 = post-breakout higher-low/20-EMA; $190 = post-print gap-fill; $278 = 6/2 spike high; $223.62 = 5/28 key-signal level
  • Sell-side won't chase: BNP $255 Neutral (6/2) and DA Davidson $250 Neutral (5/18) both below tape dispersion persists
  • Earnings blackout: defer any act within 3 trading days of confirmed Q2'26 print (~2026-08-13 est.).
  • primary (gpu-cloud dominant narrative) with squeeze overlay active since 6/1 (Jensen/short-interest) keep sizing tight even on a pullback entry; short-cover exhaustion unwinds fast.
  • Key levels: $278 = 6/2 spike high; $223.62 = 5/28 key-signal level; $210 = 20-EMA/higher-low; $190 = post-print gap-fill.
  • Pair/peer map: CRWV primary (debt-funded; margin/PT cut = leading bear tell), APLD/IREN higher-beta proxies, BE two-way power read-through, NVDA demand backdrop.
  • Balance-sheet differentiator (~$2B post-Yandex cash + $2.3B Q1 operating cash flow on prepayments, no dilution need) is the non-consensus leg vs debt-funded CRWV.
  • Sell-side sits below the tape: BNP $255 Neutral (6/2), DA Davidson $250 Neutral (5/18) dispersion persists; watch for PT upgrades chasing price as confirmation.
  • Conviction is an entry-timing call, not a thesis call: narrative ACCELERATING but extended; re-rate to HIGH on a 20-EMA pullback that holds $210 and reclaims $223.62.
  • 6/5 is the first distribution day off the $278 high the pullback the prior notes asked to wait for is now developing; do not chase the spike.

Related · shared themes