NBIS — Meta $27B Deal Analysis

wsm | 16 March 2026


The Deal

This is a transformational contract. Full stop.

$27 billion over 5 years with Meta:

Built on Nvidia Vera Rubin platform — one of the first large-scale Rubin deployments globally.

FY26 guidance unchanged — this deal starts delivering revenue in 2027, not 2026.


Why This Is Structurally Important

Let me update the contracted revenue picture:

Customer Contract Value Type Status
Microsoft $19.4B Multi-year Delivering (first tranche Nov'25)
Meta (existing) ~$3B Multi-year Servicing
Meta (new) $27B 5-year Starts early 2027
Total Named ~$49.4B

$49.4B in named contracts against a $32B market cap. That ratio has gone from 0.9x to 1.5x in a single announcement.

But here's the thing — and this is the nuance the market may be missing:

The Buyer-of-Last-Resort Structure Is Genius

The $15B tranche is structured so that Meta only buys what Nebius can't sell to other customers. In other words:

  1. Nebius builds capacity
  2. Nebius sells to multi-tenant cloud customers first (higher margin, stickier, recurring)
  3. Whatever Nebius can't fill, Meta takes at contracted prices

This is the best of both worlds. Nebius gets downside protection (guaranteed buyer for all capacity) while preserving upside (multi-tenant revenue is higher-margin and builds the durable cloud business). The bridge strategy I flagged at MS TMT is now explicitly architected into the contract structure.

If Nebius fills the capacity with cloud customers → higher revenue per GPU, higher margins, and Meta takes less. If demand softens → Meta absorbs the slack at contracted prices, protecting Nebius from overcapacity.

This eliminates the biggest risk in the neocloud model: building capacity into a demand downturn.


Impact on the Thesis

Prior Thesis (pre-deal):

Bridge strategy: hyperscaler contracts finance the buildout; multi-tenant cloud is the durable business. Risk: can multi-tenant replace hyperscaler before bridge fades?

Updated Thesis:

Bridge is now a highway. Meta deal extends hyperscaler revenue visibility to 2032 while the buyer-of-last-resort structure explicitly protects Nebius's ability to build the cloud business. The risk of "bridge fading before cloud matures" is materially reduced — Meta will absorb any gap.

What Changed:

  1. Revenue visibility: From ~22.4Bcontractedto 49.4B contracted. That's 15x FY26 guided revenue.
  2. Financing risk reduced: $27B of contractual commitments dramatically improves debt financing terms. Banks lend against contracted cash flows.
  3. Vera Rubin validation: Being one of the first large-scale Rubin deployments means Nvidia is prioritising Nebius for next-gen silicon. Combined with the $2B Nvidia investment from last week, the supply relationship is now the deepest in the neocloud space.
  4. Overcapacity risk eliminated: The buyer-of-last-resort mechanism means every GPU Nebius deploys has a guaranteed monetisation path. This was the #1 risk for all neoclouds.

Valuation Update

Metric Pre-Deal Post-Deal
Market Cap ~$25B ~$32B (+14% today)
Named Contracts $22.4B $49.4B
Contracts / Mkt Cap 0.9x 1.5x
EV/FY26 Rev ($3.2B mid) ~7.9x ~10x
EV/FY27 Rev (est $6-8B) ~4-5x
EV/FY26 Adj EBITDA ~19.9x ~25x

At ~$32B, the stock has re-rated on the deal but is NOT expensive relative to the contracted revenue trajectory. $49.4B of named contracts against a $32B market cap is a ratio I've rarely seen in growth investing. The entire current market cap is covered by named, contracted revenue.

The key question becomes execution and CapEx financing, not demand visibility. And the $2B Nvidia investment + $27B Meta commitment dramatically improve the financing environment.


What I'm Watching

  1. Q1 FY26 revenue — I said I'd add if >$600M. This deal doesn't change that trigger because it delivers in 2027, but it materially de-risks the add.
  2. CapEx financing — $16-20B FY26 CapEx, 60% self-funded. The Meta contract should help secure the 40% external financing on favourable terms.
  3. Multi-tenant mix — If the buyer-of-last-resort mechanism works as designed, the share of multi-tenant (non-hyperscaler) revenue should grow. This is the margin and durability story.
  4. CoreWeave comparison — CoreWeave just IPO'd. Nebius now has 49.4BincontractsvsCoreWeaves 15B backlog. Market cap roughly similar. NBIS has the better contract book.

Action

Hold at 10.5%. Conviction strengthening. This deal resolves the #1 risk (overcapacity/demand gap) and extends revenue visibility to 2032. The bridge-to-cloud thesis is now structurally protected by the buyer-of-last-resort mechanism.

Add trigger remains Q1 FY26 revenue >$600M — but the risk/reward of the add has improved materially. If Q1 confirms the trajectory, I'd move to 13-15%.

→ This is now the most de-risked neocloud investment I can find. $49.4B in named contracts. 70% gross margins. Nvidia as both investor and supply partner. Meta as guaranteed buyer. The execution risk is real (you still have to BUILD all this capacity), but the demand risk is effectively zero.

-wsm (Long NBIS, 10.5%)


Sources: