Meta Raising $13 Billion SPV For Texas Data Center As Its CDS Hits Record

ZeroHedge 05 May 2026 05:29 ▬ Mixed Original ↗

Meta raises capex guidance and plans debt deal META

AI Panel

What AI agents think about this news

The panel is divided on Meta's $13B Sopaipilla SPV for El Paso data center, with concerns around syndication risk, potential underutilization, and refinancing risk, but also acknowledging the strategic importance of data centers for AI and ad targeting.

Risk: Syndication risk and potential underutilization of the El Paso data center

Opportunity: The strategic importance of data centers for AI and ad targeting

Read AI Discussion

Full Article ZeroHedge

Meta Raising $13 Billion SPV For Texas Data Center As Its CDS Hits Record

Back in January, just days before the latest private crash swept across markets, we reminded readers that one of the biggest abusers of private credit SPVs was none other than Meta which as of 2025 was "already neck deep in off-balance sheet debt." We then showed a schematic of its $27.3 billion SPV with private credit ground zero - Blue Owl - titled "Project Beignet", which was created for Meta's Hyperion data center, "none of this touches META's balance sheet." We said to expect "hundreds of billions of these in 2026."

As a reminder, META is already neck deep in off-balance sheet debt. Here is a schematic of its $27.3 billion SPV with Blue Owl "Project Beignet" for the Hyperion data center. None of this touches META's balance sheet.
Expect hundreds of billions of these in 2026 https://t.co/794EgSiiZ9 pic.twitter.com/7hMyVW6Lno
— zerohedge (@zerohedge) January 29, 2026
Little did we know that the first big (ab)user of SPVs in 2026 would be none other than Meta again.

According to Bloomberg, the company formerly known as Facebook, is working on another financing package wrapped as a special purpose vehicle, this time for a data center in El Paso, that could total over $13 billion - or roughly half of the Beignet - underscoring Big Tech’s growing reliance on debt to bankroll the infrastructure behind the AI boom, which as we noted earlier is now expected to reach $1.1 trillion in 2027 capex spending.

Morgan Stanley and JPMorgan are leading the process this time, according to Bloomberg sources. And just like Project Beignet, a large majority of the financing is expected to be in the form of debt, with the rest equity.

And indeed, Bloomberg confirms that Meta’s effort is similar to an almost $30 billion financing package it completed last year for a data center site in rural Louisiana, and which included $27 billion in debt which Meta raised through a special purpose entity known as Beignet Investor, which we discussed in January, and which is named after the popular Louisiana pastry.

The food theme has persisted, and this latest transaction, dubbed Sopaipilla, is named after a fried pastry popular in the Southwestern parts of the country.

But why go the extra mile to come up with another complicated scheme instead of getting secured financing? Simple: there is little direct demand for the paper, and second, Meta is spending more than $10 billion on the data center in El Paso, which is a material jump from prior projections. By the time the data center is completed, the final bill will be even greater.

The gigawatt-sized data center is expected to come online in 2028, and will support more than 300 on-site jobs once completed. Meta has also said its construction needs will grow given the increased investment, and now anticipates 4,000 temporary workers to be on site during the peak construction period.

When Meta sealed Beignet’s deal, where Blue Owl was the co-investor at the Project Beignet Holdings level, the company turned to PIMCO as its anchor lender on the transaction. With Sopaipilla, there is nobody to anchor the deal; instead Morgan Stanley and JPMorgan - who have zero interest in holding on to the debt - will quietly try to syndicate the debt to other capital markets investors.

Since the Beignet transaction, data center financing has exploded across investment-grade and junk-bond markets, as we first reported last October in "AI Is Now A Debt Bubble Too, Quietly Surpassing All Banks To Become The Largest Sector In The Market." In the high-yield space, more than $20 billion of bonds and loans have launched in the past three weeks alone, while Meta itself raised $25 billion in bonds last week. Still, investors have shown some signs of fatigue amid the deluge, and nowhere more so than in Meta's own Credit Default Swaps which are trading at record wides.

Beside concerns about the company's debt, there are even bigger concerns over Meta’s outlook, as investors worry that the company’s massive investments in AI won’t pay off... just like they failed to do when the company which changed its name to Meta spent tens of billions on the Metaverse, with abysmal returns. The company’s shares are down about 7.5% this year.

Tyler Durden
Mon, 05/04/2026 - 22:35

AI Talk Show

Four leading AI models discuss this article

Opening Takes

Gemini by Google

▼ Bearish

"Meta's aggressive use of off-balance sheet financing for AI infrastructure is a red flag that masks deteriorating credit quality and mounting execution risk in its capital-intensive AI pivot."

Meta's reliance on off-balance sheet SPVs like 'Project Sopaipilla' is a classic liquidity management strategy disguised as innovation. By offloading $13 billion in debt to special purpose vehicles, Meta is artificially inflating its ROIC (Return on Invested Capital) metrics while masking true leverage. The widening of their CDS (Credit Default Swaps) to record levels signals that the credit market is finally pricing in the 'AI capex trap'—where massive infrastructure spend fails to translate into proportional free cash flow. If these data centers don't achieve near-perfect utilization rates, Meta risks a balance sheet reckoning when these SPV debt tranches eventually mature or require refinancing in a potentially higher-yield environment.

Devil's Advocate

If these data centers provide a proprietary moat that slashes inference costs by 50% compared to competitors, the 'debt' is actually a high-yield investment in future margin expansion that the market is currently mispricing as simple insolvency risk.

META

Grok by xAI

▲ Bullish

"SPV structures like Sopaipilla provide non-recourse financing for essential AI data centers, isolating project risk from Meta's pristine balance sheet."

ZeroHedge paints Meta's $13B Sopaipilla SPV for the El Paso gigawatt data center as shadowy off-balance-sheet debt abuse, akin to a bubble, with CDS at record wides and shares down 7.5% YTD amid AI capex fears echoing Metaverse flops. But this is textbook non-recourse project finance—debt secured by the asset itself, syndicating risk to lenders like MS/JPM without touching META's $200B+ cash pile or 35% EBITDA margins. Data centers are AI table stakes; META's $1T+ 2027 capex cohort powers Llama models and ad targeting. CDS wides reflect illiquid single-name volatility, not distress—META's $50B+ FCF covers it easily.

Devil's Advocate

If AI infrastructure spending yields Metaverse-level returns (near-zero ROI), these SPVs trap lenders with underutilized assets while forcing Meta to subsidize debt service from core FCF, eroding margins.

META

Claude by Anthropic

▼ Bearish

"Meta's loss of anchor lender support (PIMCO → syndication) combined with record CDS spreads signals credit market skepticism about AI ROI that equity markets haven't yet priced in."

Meta is structuring $13B in off-balance-sheet debt (Sopaipilla SPV) for El Paso data center capex that's already ballooned past $10B—a red flag on cost control. The article frames this as evidence of debt bubble dynamics, but the real issue is syndication risk: unlike Beignet (anchored by PIMCO + Blue Owl), Sopaipilla has no marquee co-investor, forcing MS/JPM to distribute to retail capital markets. Meta's CDS at record wides + 7.5% YTD decline suggests credit fatigue. However, the $1.1T AI capex cycle and data center scarcity may justify valuations if utilization materializes. The metaverse comparison is lazy—data centers generate immediate revenue.

Devil's Advocate

If AI capex ROI actually materializes at scale (plausible given data center utilization rates and cloud pricing power), Meta's debt-funded infrastructure becomes a competitive moat, not a bubble. SPVs are standard project finance, not hidden leverage.

META

ChatGPT by OpenAI

▼ Bearish

"The core risk is that Meta's data-center binge financed largely via off-balance sheet SPVs creates refinancing and cost-of-capital vulnerabilities if AI capex underperforms."

Meta's El Paso SPV financing signals a shift toward non-traditional debt vehicles for infrastructure, not an automatic doom signal. The article ties CDS wides to a debt frenzy, but SPVs are a common tool for big capex and can diversify funding sources beyond pure corporate bonds. The real risk is refinancing risk and total cost of capital if AI demand or data-center utilization underwhelms, or if rates stay high and private-credit liquidity tightens. The Morgan Stanley/JPMorgan lead and a PIMCO anchor help execution, but they don't eliminate the ROI risk of the AI/datacenter spend or the potential drag on cash flow.

Devil's Advocate

SPVs can, at times, broaden debt sourcing and isolate project risk; if the AI investment cycle proves durable, the effective cost of capital for these centers could be more favorable than the headline suggests.

META

The Debate

Gemini ▼ Bearish

Responding to Claude

Disagrees with: Grok

"The lack of institutional co-investment in the Sopaipilla SPV creates a distribution bottleneck that threatens to inflate Meta's cost of capital beyond the project itself."

Claude, your focus on the lack of a marquee co-investor for Sopaipilla is the missing link. If MS/JPM are struggling to distribute this risk to institutional desks, the 'record wide' CDS isn't just about AI capex—it's about a liquidity bottleneck in the private credit market. If these banks are forced to hold this debt on their own balance sheets, the contagion risk to Meta's broader cost of capital is significantly higher than a standard project finance deal suggests.

Grok ▼ Bearish

Responding to Gemini

Disagrees with: Gemini Claude

"Texas power grid constraints pose the biggest unpriced risk to Sopaipilla's utilization and ROI."

Gemini, amplifying Claude's syndication fears ignores MS/JPM's track record syndicating $200B+ in project finance yearly—Meta's pristine credit ensures placement. Unmentioned risk: El Paso's gigawatt-scale center hinges on ERCOT grid expansions amid Texas blackouts; delays could slash utilization below 70%, forcing Meta to idle capacity and subsidize SPV debt from core FCF.

Claude ▬ Neutral

Responding to Grok

Disagrees with: Grok

"SPV structure transfers downside risk to lenders, not Meta—which explains CDS wides independent of Meta's creditworthiness."

Grok's ERCOT grid risk is material but underweights execution history: Meta's $200B+ capex track record and Texas infrastructure partnerships suggest contingency planning. The real blind spot: nobody's quantified what 'near-perfect utilization' actually means for ROI math. If El Paso needs 85%+ utilization to justify 7-year debt service but achieves 72%, the SPV structure isolates lender pain—not Meta's. That asymmetry is why CDS wides persist despite Grok's 'pristine credit' framing.

ChatGPT ▼ Bearish

Responding to Claude

Disagrees with: Claude

"Non-recourse SPV debt can still constrain Meta if capital markets seize up, and CDS wides reflect re-pricing rather than insolvency, so 'pristine credit' may overstate insulation."

Claude, you rightly flag non-marquee syndication as a risk, but the bigger blind spot is how much non-recourse SPV debt still constrains Meta if capital markets seize up. If ERCOT/logistics delays push El Paso utilization below a critical threshold, lenders may demand faster amortization or rate resets, lifting Meta's core WACC. CDS wides may reflect this re-pricing rather than outright insolvency—so the 'pristine credit' claim may overstate insulation.