U.S. DOJ Moves to Ban, Agencies Rush to Buy: Claude Gets 40%-Off Procurement Deal, Usage Surges 20x

Author: TechFlow

TechFlow Editor's Note:Washington is staging an absurd policy rift. On one side, the U.S. Department of Justice is pressing hard in court to restore a government-wide ban on AI unicorn Anthropic; on the other, the General Services Administration (GSA), which oversees federal procurement nationwide, quickly announced a deepened partnership the next day.

Under the latest agreement, government agencies across the country can not only access the Claude model family at discounts of up to 40% off, but also benefit from pay-as-you-go pricing with no minimum-spend threshold. Frontline civilian agencies have effectively undercut the DOJ's hardline ban with real orders.

From removing Claude from procurement lists early this year in line with Trump's directive, to swiftly restoring and expanding purchases after a court ruled the ban unlawful, more than 500 federal and local agencies drove a 20x surge in civilian usage over the past year. This surreal reality of suing to ban while lining up to buy exposed an unbridgeable gap within the state apparatus between political posturing and essential technology demand.

Dramatic Standoff: Appeal for a Ban One Minute, 40%-Off Mega-Deal the Next

Washington's bureaucracy has displayed a jaw-dropping double standard toward major model developers. On Oct. 26, the U.S. Department of Justice filed a brief with the U.S. Court of Appeals for the Federal Circuit, vowing to overturn the prior ruling through the courts and restore a government-wide ban on AI leader Anthropic.

Ironically, just one day later, the General Services Administration (GSA), the authoritative agency overseeing federal administrative assets and procurement lists nationwide, sent a sharply opposite signal by publicly releasing a new framework agreement to expand strategic procurement with Anthropic.

According to theAnthropic official announcement,the disclosed commercial terms show that the partnership enables government agencies at all levels across the United States to access Claude's core model family with discounts of up to 40% (i.e., 60% of list price), and billing has fully shifted to settlement based on actual Token consumption, with no punitive minimum-spend commitments required.

The DOJ also accused Anthropic's models in court of potentially endangering government data security and sovereign boundaries, while GSA, which controls the purse strings and procurement approvals, proactively secured bulk-purchase discounts for major departments. Such contradictory administrative disconnect has stunned Washington policy observers.

Political Compliance U-Turn: From Enforcing a Ban to Court-Ordered Correction

GSA's current enthusiastic embrace of Claude contrasts sharply with the sweeping political purge earlier this year. Looking back at the procurement compliance saga, GSA executed several dramatic 180-degree reversals in a short period, reflecting how executive orders repeatedly hit the wall within the rule-of-law framework.

In February this year, after the Trump team publicly called for severe punishment and a ban on Anthropic, GSA immediately showed high political-compliance sensitivity and promptly removed Claude entirely from the government procurement catalog. Agencies were then required to stop using related services, causing significant office disruption.

However, the administrative ban soon suffered a legal setback. In March, a U.S. district court in California issued an emergency halt to the White House's comprehensive ban directive; shortly afterward in April, GSA shifted course and swiftly restored Anthropic's eligibility to bid for federal government procurement, setting the stage for subsequent cooperation.

By August this year, a federal judge delivered a final ruling, clearly determining that the federal government's broad ban on Anthropic constituted an administrative violation. Although the DOJ has firmly appealed, the ruling remains legally effective during appellate proceedings and cannot prevent GSA from advancing the discounted agreement.

Essential Demand Cannot Be Stopped: 500 Government Agencies Join, Civilian Usage Surges 20x

Although political litigation continues fiercely, large-model adoption in the day-to-day operations of federal agencies has already crossed an irreversible tipping point. Facing routine tasks such as document summarization, policy drafting, and data comparison, the massive government system has become deeply dependent on advanced AI.

According to the latest operational data disclosed by Anthropic, approximately 500 federal, state, and local government agencies across the United States are now deeply using Claude in their daily work. From frontline environmental and transportation inspection departments to state-level judicial support systems, large models have become bona fide foundational productivity tools.

Even more striking is the slope of growth: in just the past year, actual Claude call volume across U.S. federal civilian agencies has exploded exponentially, with overall usage surging by about 20 times. This intense bottom-up demand driven by operational units has completely hollowed out the ban attempts by senior politicians.

For agency heads facing budget cuts and massive backlogs of government work, Claude's 40% discount and zero minimum-spend threshold are a timely lifeline for cost reduction and efficiency gains. Even with the Department of Justice's big stick hanging overhead, frontline departments will still cast a vote of confidence in the technology's usability through concrete procurement budgets.

Cognitive rift: systemic failure of ideological blockades amid the general-purpose technology wave

This farce of 'ban first, embrace later' reveals an unavoidable objective rule to the global technology and policy communities: once artificial general intelligence (AGI) infrastructure is deeply embedded in the functioning of modern society, any ban directives driven by partisan self-interest or short-term political ideology will fail completely.

The Department of Justice represents a reactive defense by the traditional bureaucracy under geopolitical anxiety and directives from political strongmen, trying to build a wall with a one-size-fits-all ban; but the GSA and hundreds of government agencies represent highly pragmatic demands for governance efficiency. Their head-on clash exposes the intense backlash that a technological surge has triggered against traditional governance.

When a productivity tool can multiply document-processing efficiency by several times and genuinely save tens of millions of dollars in fiscal budgets, the self-sustaining logic of administrative agencies instinctively bypasses blockades. The emergence of the 40% discount agreement proves that as long as the technology itself is sufficiently advanced and its value for money is extreme enough, administrative barriers must ultimately yield to efficiency.

As Anthropic becomes more deeply tied to the federal government procurement network, the Department of Justice's future appeals litigation will inevitably become even more awkward. This internal tug-of-war within the bureaucracy will not only fail to prevent the ubiquitous adoption of advanced large models, but will instead show the world the tenacious resilience of technological pragmatism amid political noise.

Source: 深潮 TechFlow