CoreWeave, the NVIDIA-backed AI cloud infrastructure company that went public just one year ago, reported full-year 2025 revenue of $5.13 billion – a staggering 168% increase year-over-year – while simultaneously announcing plans to spend upwards of $30 billion in capital expenditure during 2026. The company’s Q4 2025 earnings, released in early March 2026, revealed a business growing at breakneck speed but burning through cash at a rate that has split Wall Street between bulls who see the next great cloud platform and bears warning of an overleveraged house of cards built on GPU debt.

With a revenue backlog of $66.8 billion, total liabilities approaching $29 billion, and a stock that has tripled since its March 2025 IPO at $40 per share, CoreWeave has become the most polarizing company in the AI infrastructure market. The question facing investors, customers, and the broader technology industry in March 2026 is whether CoreWeave’s massive bet on AI compute will pay off – or whether the company’s debt-fueled growth trajectory is unsustainable.

CoreWeave’s Q4 2025 Earnings: Record Revenue Meets Record Losses

CoreWeave’s fourth-quarter 2025 results painted a picture of extraordinary growth accompanied by mounting losses. The company reported Q4 revenue of $1.57 billion, more than doubling the $747 million recorded in Q4 2024 – a 110% year-over-year increase that exceeded most analyst expectations. For the full fiscal year 2025, total revenue reached $5.13 billion, up from $1.92 billion in 2024, representing 168% annual growth.

However, the bottom line told a different story. CoreWeave posted a net loss of $1.17 billion for the full year, compared to $863 million in 2024. Q4 alone saw an adjusted net loss of $284 million, with earnings per share of -$0.89 – significantly missing the analyst consensus estimate of -$0.21. The widening gap between revenue growth and profitability has become the central debate among CoreWeave analysts.

Adjusted EBITDA offered a more optimistic view, reaching $898 million for the quarter with a 57% margin, and $3.09 billion for the full year, up from $1.22 billion in 2024. Adjusted operating income, however, declined to $88 million in Q4 (a 6% operating margin), down from $121 million in the same period a year earlier – a troubling signal that scaling costs are outpacing revenue gains at the operating level.

“CoreWeave’s top-line growth is undeniable, but the market is right to scrutinize the path to profitability,” said Dan Ives, Senior Equity Research Analyst at Wedbush Securities. “At some point, the revenue story needs to translate into free cash flow, and that timeline keeps getting pushed further out with each new round of capex guidance.”

The $66.8 Billion Backlog: A Revenue Pipeline or a Mirage?

Perhaps the most eye-catching figure in CoreWeave’s earnings report was its revenue backlog, which reached $66.8 billion as of December 31, 2025. This figure represents remaining performance obligations and estimated future revenue from committed customer contracts – and it nearly doubled from approximately $35 billion in the prior quarter. The backlog suggests that demand for CoreWeave’s AI compute services is accelerating, with enterprise and hyperscaler customers locking in long-term capacity commitments.

Customer agreements signed in recent periods represented a cumulative $7.8 billion in committed spend over 12-month terms from signing, with those agreements showing approximately 4x growth on initial contract value over their respective periods. This expansion rate indicates that customers who start with CoreWeave tend to scale their usage dramatically – a strong signal for future revenue realization.

However, skeptics point out that backlog figures are forward-looking projections, not guaranteed revenue. Contract cancellations, renegotiations, or shifts in AI spending priorities could materially reduce the actual revenue recognized from this pipeline. The concentration of CoreWeave’s customer base adds another layer of risk – while the company has not disclosed exact percentages, industry analysts estimate that a small number of hyperscaler clients, including Microsoft, account for a disproportionate share of total revenue.

“A $67 billion backlog is impressive on paper, but investors need to understand how much of that is truly contracted versus estimated,” noted Patrick Moorhead, CEO and Chief Analyst at Moor Insights & Strategy. “In the cloud infrastructure market, contract structures can change rapidly, especially if the AI spending cycle decelerates.”

The $30 Billion Capex Gamble: Building the AI Cloud at All Costs

CoreWeave’s most controversial decision is its planned capital expenditure for 2026, which management has guided at $30 billion or more. To put this figure in perspective, CoreWeave deployed $8.2 billion in capex during Q4 2025 alone – more than many large enterprises spend in an entire year. The company’s infrastructure buildout is focused on deploying next-generation NVIDIA GPUs across an expanding network of data centers to meet the surging demand for AI training and inference compute.

This level of spending would place CoreWeave’s 2026 capex in the same tier as companies like Meta, which committed roughly $60-65 billion to AI infrastructure in 2025, and well ahead of most pure-play cloud providers. The critical difference is that Meta generates over $150 billion in annual advertising revenue to fund its infrastructure investments, while CoreWeave’s $5.13 billion revenue base must support a capex plan that is nearly six times larger.

The company has been aggressively raising capital to fund this expansion. During Q4 2025, CoreWeave completed a $2.6 billion convertible senior notes offering and expanded its revolving credit facility to $2.5 billion. Earlier reports indicated the company raised approximately $8.5 billion from banks and investors for cloud expansion, contributing to total liabilities that have ballooned to approximately $29 billion against just $3.9 billion in equity.

Interest expense has become a significant and growing burden. In Q3 2025, interest costs hit $310.6 million – three times the $104.4 million recorded in Q3 2024 – representing 23% of quarterly revenue. With 2031 bonds yielding 11.5%, the market is pricing in substantial credit risk for CoreWeave’s debt.

CoreWeave’s Financial Performance: Key Metrics at a Glance

Metric	Q4 2024	Q4 2025	FY 2024	FY 2025	Change (YoY)
Revenue	$747M	$1.57B	$1.92B	$5.13B	+168%
Net Loss	N/A	-$284M (adj.)	-$863M	-$1.17B	-35.6%
Adjusted EBITDA	N/A	$898M	$1.22B	$3.09B	+153%
EBITDA Margin	N/A	57%	63.5%	60.2%	-3.3pp
Adj. Operating Income	$121M	$88M	$324M	N/A	-27.3%
Revenue Backlog	N/A	$66.8B	N/A	$66.8B	~2x QoQ
Total Liabilities	N/A	~$29B	N/A	~$29B	N/A
Q4 Capex	N/A	$8.2B	N/A	N/A	N/A

From Crypto Mining to AI Cloud: CoreWeave’s Unlikely Origin Story

CoreWeave’s trajectory from a cryptocurrency mining operation to the most talked-about AI infrastructure company on Wall Street is one of the more remarkable pivots in recent technology history. Founded in 2017 by Michael Intrator, Brian Venturo, and Brannin McInley, the company originally operated Ethereum mining rigs, using GPU hardware to mine cryptocurrency. When Ethereum transitioned to a proof-of-stake consensus mechanism in 2022, eliminating the need for GPU-based mining, the founders recognized that their expertise in managing large-scale GPU infrastructure could serve the rapidly growing AI market.

The pivot proved prescient. By 2024, CoreWeave had raised $3.89 billion in private funding, including a $1.10 billion Series C round in April 2024. The company’s March 2025 IPO on the Nasdaq Global Select Market, priced at $40 per share, raised approximately $1.5 billion in gross proceeds. Morgan Stanley, J.P. Morgan, and Goldman Sachs served as lead underwriters – a blue-chip roster that signaled Wall Street’s confidence in the AI infrastructure thesis.

Pre-IPO private market trading on secondary platforms had valued CoreWeave stock at $74.24 per share, nearly double the eventual IPO price – a gap that raised questions about whether the company had deliberately priced conservatively to ensure a successful debut. In the year since going public, the stock has more than tripled, validating the early investors who saw CoreWeave as a pure-play bet on AI compute demand.

The NVIDIA Relationship: CoreWeave’s Greatest Asset and Biggest Risk

CoreWeave’s relationship with NVIDIA is both its most valuable strategic asset and a significant source of concentration risk. NVIDIA is not only CoreWeave’s primary GPU supplier but also a shareholder in the company – a dual relationship that gives CoreWeave preferential access to the most sought-after AI chips in the world, including the latest Blackwell B200 and B300 GPUs.

This relationship has been a key competitive differentiator. While traditional cloud providers like AWS, Azure, and Google Cloud also deploy NVIDIA hardware, CoreWeave’s purpose-built infrastructure and close partnership with NVIDIA have allowed it to offer specialized AI compute services with lower latency and higher GPU utilization rates. Enterprise customers running large-scale AI training workloads have increasingly turned to CoreWeave as an alternative to the hyperscaler cloud platforms.

However, the dependency on NVIDIA creates vulnerability. Any disruption to the GPU supply chain, changes in NVIDIA’s pricing or allocation strategy, or the emergence of competitive chip architectures from AMD, Intel, or custom silicon efforts by companies like OpenAI could impact CoreWeave’s cost structure and competitive positioning. The company’s massive capex plans assume continued access to next-generation NVIDIA hardware at favorable terms – an assumption that becomes riskier as more players compete for limited GPU supply.

“CoreWeave’s NVIDIA relationship is a double-edged sword,” observed Stacy Rasgon, Senior Semiconductor Analyst at Bernstein. “It gives them access that others don’t have, but it also means their entire business model is built on a single supplier’s roadmap. If NVIDIA’s pricing power increases—which it has every incentive to do—CoreWeave’s margins could compress significantly.”

Wall Street Divided: Analyst Ratings and Price Targets

The analyst community remains broadly bullish on CoreWeave, though with significant caveats about valuation and risk. As of March 2026, the stock carried Buy ratings from several major firms, including Jefferies, Roth Capital, Freedom Capital Markets, HC Wainwright & Co., and Compass Point. Wells Fargo maintained an Overweight rating, while Evercore ISI Group rated the stock as Outperform.

The bull case centers on CoreWeave’s dominant position in the fast-growing AI cloud infrastructure market, its $66.8 billion revenue backlog, and the structural tailwinds driving enterprise AI adoption. Proponents argue that CoreWeave is building the computing infrastructure that will underpin the next generation of AI applications, and that the current investment cycle – while expensive – will generate enormous returns as the backlog converts to recognized revenue.

The bear case focuses on CoreWeave’s use, profitability timeline, and customer concentration. With $29 billion in total liabilities against $3.9 billion in equity, the company’s balance sheet is among the most leveraged in the technology sector. Rising interest expense – which tripled year-over-year to $310.6 million in Q3 2025 – threatens to consume an ever-larger share of revenue. And the prospect of $30 billion or more in 2026 capex means the debt pile will continue growing before the company can realistically achieve free cash flow breakeven.

“Rising debt would drag interest expense from large to structural, and eventually force the business to run for its lenders before it can run for shareholders,” warned one widely cited analysis on the company’s financial trajectory. CoreWeave’s 2031 bonds yielding 11.5% reflect the market’s assessment of substantial credit risk despite the company’s rapid revenue growth.

CoreWeave Stock Performance Since IPO

CoreWeave’s stock has been among the most volatile in the technology sector since its March 2025 listing. After pricing at $40, shares initially traded sideways before embarking on a dramatic rally that took the stock above $187 at its 52-week high. By late March 2026, shares had pulled back significantly from those peaks, settling around $73 after a turbulent final week of the month.

Date	Close Price	Daily Change	Volume
March 4, 2026	$79.50	+2.19%	25.3M
March 5, 2026	$74.82	-2.83%	22.7M
March 24, 2026	$83.02	+0.11%	21.6M
March 25, 2026	$87.58	+2.76%	19.1M
March 26, 2026	$80.45	-7.37%	18.5M
March 27, 2026	$74.81	-4.59%	22.2M
March 30, 2026	$69.15	-7.95%	25.1M
March 31, 2026	$73.45	+0.62%	3.1M

The late-March selloff saw CoreWeave shares drop from $87.58 on March 25 to $69.15 on March 30 – a decline of more than 21% in just three trading sessions. The selloff coincided with broader market weakness in technology stocks and heightened investor concerns about the company’s capex guidance and path to profitability. Daily trading volumes regularly exceeded 20 million shares, reflecting intense institutional activity. Over the trailing 12 months, CoreWeave stock delivered a 38.69% total return – impressive, but well below the more than 200% gain from IPO price to its all-time high.

The Competitive Landscape: Neoclouds vs. Hyperscalers

CoreWeave operates in an increasingly crowded AI cloud infrastructure market that includes both traditional hyperscaler cloud providers and a growing cohort of specialized “neocloud” competitors. Understanding where CoreWeave fits in this landscape is essential for evaluating its long-term prospects.

Hyperscaler Competition

Amazon Web Services, Microsoft Azure, and Google Cloud Platform collectively dominate the global cloud infrastructure market and have each committed tens of billions of dollars to AI-specific compute capacity. AWS has expanded its custom Trainium chip offerings, Microsoft has deepened its OpenAI partnership to secure exclusive AI compute, and Google has continued developing its TPU architecture. These platforms offer AI compute alongside their full suite of cloud services, providing a one-stop-shop value proposition that pure-play GPU cloud providers cannot match.

CoreWeave’s advantage over hyperscalers lies in specialization. By focusing exclusively on GPU compute, CoreWeave can offer purpose-built infrastructure with optimized networking, storage, and orchestration for AI workloads. Enterprise customers running large-scale model training often find that CoreWeave’s specialized environment delivers better performance per dollar than general-purpose cloud platforms.

Neocloud Rivals

The neocloud category includes several well-funded competitors targeting the same AI compute market. Lambda Labs has built a significant GPU cloud business and raised substantial private capital. Crusoe Energy has differentiated itself by powering its data centers with stranded natural gas and renewable energy, appealing to customers with sustainability requirements. Together AI has focused on AI inference optimization, while Cerebras offers its own custom wafer-scale AI chip as an alternative to NVIDIA GPUs.

“The AI infrastructure market is big enough for multiple winners, but CoreWeave’s scale advantage is real,” said Matt Garman, former Senior Vice President at AWS, in a recent industry conference. “The question is whether that scale can be sustained without the diversified revenue streams that hyperscalers rely on to weather cycles.”

The Debt Dilemma: Can CoreWeave Outgrow Its Use?

CoreWeave’s financial structure represents one of the most aggressive debt-fueled growth strategies in modern technology history. With total liabilities of approximately $29 billion against $3.9 billion in equity, the company’s debt-to-equity ratio exceeds 7:1 – a level that would be considered extreme in almost any industry.

The company’s debt service obligations are equally daunting. With interest expense running at over $300 million per quarter and climbing, CoreWeave is spending roughly 20-25% of its revenue on interest payments alone. The 11.5% yield on its 2031 bonds indicates that credit markets view CoreWeave’s debt as high-yield – essentially junk bond territory – despite the company’s rapid revenue growth.

The bull argument is that CoreWeave’s $66.8 billion backlog provides sufficient revenue visibility to service and ultimately retire this debt. If the company can convert even half of its backlog into recognized revenue over the next 3-5 years, it would generate enough cash flow to deleverage naturally while continuing to invest in growth. The bears counter that backlog conversion rates are uncertain, customer concentration creates binary risk, and the AI infrastructure spending cycle could decelerate before CoreWeave achieves breakeven.

Historical parallels offer mixed guidance. Companies like Level 3 Communications and Global Crossing built massive fiber-optic infrastructure during the late 1990s telecom boom, accumulating enormous debt in the process. When demand failed to meet projections, both companies faced severe financial distress. On the other hand, Amazon Web Services itself was initially funded through years of losses and heavy investment before becoming the most profitable division of the world’s largest retailer.

“The comparison to the telecom bubble is superficial but not entirely wrong,” commented Lisa Su, CEO of AMD, during a semiconductor industry panel in February 2026. “The difference is that AI compute demand appears more structurally durable than dot-com era bandwidth demand. But the companies building this infrastructure still need to prove they can convert capacity into profitable revenue.”

The AI Infrastructure Spending Cycle: Is It Sustainable?

CoreWeave’s fate is inextricably linked to the broader AI infrastructure spending cycle. In 2025, the major technology companies collectively committed over $700 billion to AI infrastructure, including data centers, chips, networking equipment, and power generation. This spending frenzy has been driven by the explosive growth of large language models, enterprise AI adoption, and the race to build artificial general intelligence.

For CoreWeave, the sustainability of this cycle is existential. The company’s business model depends on continued demand for GPU compute at current or higher price points. If AI spending plateaus or major customers shift workloads to in-house infrastructure or alternative chip architectures, CoreWeave’s revenue growth could decelerate rapidly while its fixed debt obligations remain unchanged.

Several factors support continued AI infrastructure investment. Enterprise AI adoption is still in its early stages, with most large organizations running pilot projects rather than production-scale deployments. The emergence of agentic AI systems that can autonomously execute complex tasks is expected to drive a new wave of compute demand. And the shift from AI training to inference workloads suggests that GPU demand will become more continuous and predictable over time.

However, countervailing forces are emerging. The AI data center power crisis is constraining the pace at which new compute capacity can be brought online. Research advances in model efficiency – including mixture-of-experts architectures, quantization techniques, and smaller specialized models – could reduce the raw compute required per unit of AI capability. And the entry of custom silicon from cloud providers and AI companies could fragment demand away from NVIDIA GPUs, undermining CoreWeave’s core value proposition.

Customer Dynamics: The Microsoft Factor and Beyond

CoreWeave’s customer base includes some of the largest and most influential companies in the AI ecosystem. Microsoft, Meta, and OpenAI are all reported to be significant customers, using CoreWeave’s GPU infrastructure to supplement their own data center capacity. The presence of these marquee clients validates CoreWeave’s technical capabilities but also creates concentration risk that investors are increasingly scrutinizing.

While CoreWeave has not disclosed the exact revenue contribution from individual customers, its SEC filings indicate that no single customer accounts for more than 35% of the $66.8 billion backlog – a threshold that still implies significant reliance on a handful of large accounts. If any of these customers reduce their commitments, develop competing in-house capabilities, or negotiate more favorable pricing terms, the impact on CoreWeave’s revenue could be material.

Microsoft’s relationship with CoreWeave is particularly complex. As both a major customer and a competitor (via Azure), Microsoft’s willingness to continue outsourcing AI compute to CoreWeave will depend on the relative economics and capacity availability of its own infrastructure. As Microsoft’s AI capex spending reaches new heights, the company may eventually bring more workloads in-house, reducing its dependence on third-party providers.

Five Predictions for CoreWeave’s Future

Based on current trajectory and market dynamics, here are five predictions for CoreWeave’s evolution over the next 12-18 months:

1. Revenue will exceed $12 billion in 2026. With a $66.8 billion backlog and 168% growth rate, CoreWeave is positioned to at least double its top line. The conversion of existing contracts and new customer wins should drive revenue well above $10 billion, with $12-15 billion being a realistic range.

2. The company will pursue additional equity and debt offerings. With $30 billion in planned capex and a balance sheet already stretched to its limits, CoreWeave will need to raise significant additional capital during 2026. Expect at least one secondary stock offering and multiple debt tranches to fund the infrastructure buildout.

3. Customer diversification will become a strategic priority. To reduce concentration risk and support its valuation narrative, CoreWeave will aggressively pursue mid-market enterprise customers and international expansion. The company may acquire smaller cloud or managed services providers to accelerate this diversification.

4. A major partnership or acquisition will reshape the competitive landscape. CoreWeave’s scale and NVIDIA relationship make it an attractive target or partner for companies seeking to enter or expand in the AI infrastructure market. A strategic investment from a sovereign wealth fund, a partnership with a telecom provider, or even an acquisition by a hyperscaler is plausible within the next 18 months.

5. Profitability will remain elusive through 2027. The combination of massive capex, rising interest expense, and the lead time between infrastructure deployment and revenue recognition means that CoreWeave is unlikely to achieve GAAP profitability before 2028 at the earliest. Free cash flow breakeven may come sooner if backlog conversion rates remain strong, but the path to net income is measured in years, not quarters.

What CoreWeave’s Rise Means for the Broader Tech Industry

CoreWeave’s emergence as a major player in AI infrastructure carries implications that extend far beyond its own financial performance. The company’s success has validated the neocloud business model, encouraging other specialized cloud providers to pursue similar strategies. It has also demonstrated that the AI compute market is large enough to support significant new entrants alongside the established hyperscaler platforms.

For NVIDIA, CoreWeave’s growth is a powerful proof point for the GPU ecosystem. Every dollar CoreWeave spends on capex translates primarily into NVIDIA revenue, making CoreWeave one of NVIDIA’s most important customers. The symbiotic relationship between the two companies – with NVIDIA as both supplier and investor – has become a template for how chip companies and cloud providers can co-invest in the AI infrastructure buildout.

For enterprise customers, CoreWeave’s presence in the market creates more options and competitive pressure in AI cloud pricing. As CoreWeave and other neoclouds compete with AWS, Azure, and GCP for AI workloads, customers benefit from greater choice, potentially lower prices, and more specialized services tailored to AI workflows.

For the financial markets, CoreWeave represents both the opportunity and the risk of the AI infrastructure investment cycle. If the company succeeds, it will validate the thesis that AI demand can support massive infrastructure investments financed primarily through debt. If it falters, it could trigger a reassessment of AI infrastructure valuations across the entire technology sector – a scenario with implications for NVIDIA’s GPU pricing power, hyperscaler capex commitments, and the broader AI market narrative.

Related Coverage

More on AI Infrastructure and Cloud Computing

• Big Tech’s $700 Billion AI Infrastructure Bet: Inside the 2026 Spending Race
• The AI Data Center Power Crisis: How Big Tech’s 125 GW Appetite Is Reshaping the US Energy Grid
• NVIDIA Blackwell GPU Pricing: B200, B300 and DGX Cost Breakdown
• Microsoft’s $150 Billion AI Capex Gamble: Inside the Azure Surge and Copilot’s Adoption Crisis
• AWS vs Azure vs Google Cloud 2026: The Leading Cloud Platform Comparison
• Agentic AI in Enterprise 2026: Inside the $9 Billion Market Reshaping How Businesses Operate
• SoftBank’s $40 Billion Loan for OpenAI: Inside the Largest AI-Focused Corporate Loan in History

Frequently Asked Questions

What is CoreWeave and what does the company do?

CoreWeave is a specialized AI cloud infrastructure company that provides GPU-based compute services to enterprises, AI companies, and hyperscalers. Founded in 2017 as a cryptocurrency mining operation, the company pivoted to AI cloud computing and now operates one of the largest independent GPU cloud platforms in the world, powered primarily by NVIDIA hardware.

How much revenue did CoreWeave generate in 2025?

CoreWeave reported full-year 2025 revenue of $5.13 billion, representing 168% year-over-year growth from $1.92 billion in 2024. Q4 2025 revenue alone was $1.57 billion, more than doubling the $747 million recorded in Q4 2024.

Why is CoreWeave losing money despite rapid revenue growth?

CoreWeave posted a net loss of $1.17 billion in 2025 primarily due to massive capital expenditures on GPU infrastructure, depreciation of that equipment, and rapidly rising interest expenses on its approximately $29 billion in total liabilities. The company is investing aggressively ahead of revenue, deploying $8.2 billion in capex in Q4 2025 alone.

What is CoreWeave’s relationship with NVIDIA?

NVIDIA is both CoreWeave’s primary GPU supplier and an investor in the company. This dual relationship gives CoreWeave preferential access to NVIDIA’s latest AI chips, including Blackwell-generation GPUs. However, it also creates significant supplier concentration risk, as CoreWeave’s entire infrastructure depends on continued access to NVIDIA hardware at favorable terms.

Is CoreWeave stock a good investment in 2026?

Analyst opinion is divided. Multiple firms including Jefferies, Wells Fargo, Evercore ISI, and HC Wainwright maintain Buy or Overweight ratings, citing CoreWeave’s $66.8 billion revenue backlog and dominant AI cloud position. Bears point to $29 billion in liabilities, rising interest costs, customer concentration risk, and a stock that has already tripled from its IPO price. Investors should carefully evaluate their risk tolerance before taking a position.

How does CoreWeave compete with AWS, Azure, and Google Cloud?

CoreWeave differentiates through specialization rather than breadth. While hyperscalers offer thorough cloud platforms with AI as one of many services, CoreWeave focuses exclusively on GPU compute for AI workloads. This allows the company to offer purpose-built infrastructure with optimized networking and storage, often delivering better performance per dollar for large-scale AI training and inference tasks.

What is CoreWeave’s $66.8 billion revenue backlog?

The $66.8 billion backlog represents remaining performance obligations and estimated future revenue from committed customer contracts as of December 31, 2025. It nearly doubled from approximately $35 billion in the prior quarter, reflecting accelerating demand for CoreWeave’s services. However, backlog figures are projections and not guaranteed revenue – contract modifications, cancellations, or market shifts could affect actual revenue recognition.

Could CoreWeave go bankrupt?

While bankruptcy is not imminent given CoreWeave’s strong revenue growth and substantial backlog, the risk is non-trivial. With $29 billion in total liabilities, interest costs exceeding $300 million per quarter, and plans for $30 billion in additional capex in 2026, the company’s financial structure leaves little margin for error. A significant slowdown in AI spending, loss of a major customer, or inability to refinance debt at acceptable terms could create severe financial stress.