Executives at artificial intelligence startup Anthropic have warned that a move by the U.S. government to place the company on a national security blacklist could slash billions of dollars from its projected 2026 revenue and damage its reputation with customers and investors.

The warnings emerged in federal court filings tied to a lawsuit the company filed Monday seeking to block the U.S. Department of Defense from placing it on the blacklist. The legal challenge marks a dramatic escalation in a dispute between the AI firm and the Pentagon over restrictions governing how Anthropic’s technology can be used in military-related work.

Senior executives described the potential consequences in sworn statements submitted to the court, arguing that the designation could destabilize the company’s commercial and government-facing businesses.

Krishna Rao, the company’s chief financial officer, said the impact could reach into the billions.

“Across Anthropic’s entire business, and adjusting for how likely any given customer is to take a maximal reading, the government’s actions could reduce Anthropic’s 2026 revenue by multiple billions of dollars,” Rao said in the filing.

Rao warned that if the government’s actions are allowed to stand, the consequences for the company would be “almost impossible to reverse.”

According to the filing, Anthropic estimates that hundreds of millions of dollars in revenue tied specifically to work with the Defense Department could be at risk in 2026 alone. The company also fears broader ripple effects across the defense technology ecosystem.

Rao said Anthropic could lose between 50% and 100% of revenue linked to defense contractors and other companies whose operations depend heavily on Pentagon relationships.

Beyond direct contract losses, the company argues that the designation could undermine investor confidence at a time when AI firms are spending heavily on computing infrastructure and research.

Rao said the move would likely increase the cost of raising capital needed to fund operations and growth.

Anthropic’s Head of Public Sector, Thiyagu Ramasamy, described what he called immediate damage to the company’s standing in the government technology market.

“The government’s actions immediately and irreparably harm Anthropic,” Ramasamy said in the court filing.

“The designation also impugns Anthropic’s integrity and reputation as a trusted partner, having a real but incalculable effect on sales to non-governmental customers.”

According to Ramasamy, the company expects an immediate loss of more than $150 million in annual recurring revenue tied to current and anticipated Defense Department contracts.

The company’s public sector business had been expanding rapidly before the dispute. Ramasamy said that between December 2025 and January 2026, Anthropic recorded a fourfold increase in its annual recurring revenue run rate from government clients.

Internal projections suggested that over the next five years, the public sector business could generate revenue in the multiple billions of dollars.

Those projections now appear uncertain.

If defense contractors distance themselves from the company to avoid potential compliance risks, Ramasamy warned that Anthropic’s expected public sector annual recurring revenue — projected to exceed $500 million in 2026 — could “shrink substantially or disappear altogether.”

The company’s commercial relationships are also showing signs of strain.

Paul Smith, Anthropic’s chief commercial officer, said the controversy has already disrupted negotiations and prompted some clients to reconsider their partnerships.

In one case, a partner with a multi-million-dollar annual contract switched from Anthropic’s Claude AI system to a competing generative AI model for a deployment at the U.S. Food and Drug Administration. Smith said the move eliminated an anticipated revenue pipeline worth more than $100 million.

Other deals have also been affected.

Smith said negotiations with financial institutions representing roughly $180 million in potential contracts have been disrupted, while a $15 million agreement was paused.

One fintech customer also cut the value of an existing contract in half — from $10 million to $5 million — citing concerns over the company’s dispute with the Pentagon.

According to Smith, enterprise customers are increasingly anxious about the implications of the government’s actions.

Anthropic has received inquiries from more than 100 corporate clients expressing what he described as “deep fear, confusion and doubt” about the risks of continuing to work with the company.

The case highlights the growing tension between rapidly expanding AI firms and government efforts to regulate how advanced technology can be used in national security contexts. Companies developing powerful AI models are increasingly pursuing contracts with government agencies and defense contractors, seeing the public sector as a major growth market.

But that relationship also exposes them to heightened scrutiny from regulators and national security officials concerned about how the technology is deployed.

Goldman Sachs (GS.N) is actively marketing a derivative product to hedge funds that enables them to take long or short positions on corporate loans, with a particular focus on debt issued by software companies, according to a source familiar with the matter who spoke to the Financial Times.

The instrument — a total return swap (TRS) — allows investors to gain synthetic exposure to the performance of underlying corporate loans without owning them directly. Under a TRS, one party (typically the hedge fund) receives the total economic return of the reference asset (interest payments plus any price appreciation or depreciation), while paying a financing cost (usually based on SOFR or LIBOR plus a spread) to Goldman. The structure can be used to express bullish views (long the loans) or bearish views (short the loans) on credit quality and pricing.

No trades have been executed using this specific strategy to date, the source said.

Goldman Sachs declined to comment on the specifics of the offering but provided a general statement.

“As a market-maker, we obviously engage constantly with clients on facilitating the trading strategies they want to execute. This happens every day, across many asset classes, in every market environment,” the Wall Street giant said.

The pitch comes at a moment of acute stress in the software sector. Software-as-a-service (SaaS) and legacy enterprise-software companies have seen their stock prices and credit spreads widen significantly in 2026 as investors grow increasingly concerned that rapid advances in generative AI and agentic systems could automate or commoditize many core software functions. AI agents capable of performing complex multi-step tasks across applications — from email management and data analysis to code generation and customer support — are viewed as existential threats to traditional SaaS growth models.

The primary leveraged-loan and high-yield bond markets for software issuers have been effectively frozen. No major debt deals backed by software companies have priced in the primary market since Oracle’s $25 billion debt package closed on February 2, 2026. Secondary-market liquidity has thinned, with loan prices for many software borrowers trading at deep discounts to par amid fears of slower growth, margin compression, and potential rating downgrades.

Goldman’s TRS offering would allow sophisticated investors to express directional views on software-loan performance without needing to source physical paper in a constrained market. A short position via TRS would profit if loan prices fall further (due to credit deterioration or forced selling), while a long position would benefit from any stabilization or recovery in valuations. The product could also serve as a hedging tool for banks and other loan holders looking to mitigate downside risk.

The software sector’s credit and equity weakness is seen as part of a broader re-rating of growth stocks in the AI era. Multiples across enterprise software have compressed sharply since late 2025, with many names trading at single-digit forward EV/Revenue or EV/EBITDA levels — a stark contrast to the 15–30x multiples seen during the 2020–2022 boom.

Investors worry that AI agents could erode subscription-based recurring revenue by offering low-cost or free alternatives to legacy SaaS tools. At the same time, leveraged-loan and high-yield spreads for software issuers have widened considerably. Secondary loan prices for many B-rated software borrowers are now trading in the mid-80s to low-90s, underlining both macro caution (higher rates, slower growth) and sector-specific AI disruption risk.

Goldman’s move to facilitate TRS trading in this space underscores the Street’s view that the software loan market — once seen as a stable, high-conviction asset class — is now a fertile ground for relative-value and directional trades. The bank’s role as a leading arranger and market-maker in leveraged loans gives it unique visibility into pricing dislocations and hedging demand.

While no trades have yet been executed, Goldman’s active pitching suggests growing hedge-fund interest in expressing bearish or hedging views on software credit. If AI disruption fears intensify, or if primary issuance remains sidelined, secondary-market volatility could increase, creating more opportunities for TRS-based strategies.

For software borrowers, the lack of primary-market access raises refinancing risks, particularly for companies with near-term maturities or covenant pressure. But for investors, Goldman’s product offers a way to monetize or hedge those risks without needing to source illiquid physical loans.

The development highlights how quickly market dynamics are shifting in the AI era. What was once considered a defensive, cash-flow-rich sector (enterprise software) is now viewed by some as structurally vulnerable. Goldman’s facilitation of TRS trading on software loans is an early sign that Wall Street is adapting to this new reality — turning credit dislocation into a tradable opportunity while the primary market remains closed.