Editorial note: GadgetBond generally avoids political coverage. When technology and gadgets intersect with politics or major current events, however, we may cover the story when the technology angle is relevant to our readers. Read our statement

Elon Musk and Donald Trump have traded barbs in public for months, but on the business-technology beat, things are stranger than a Twitter spat: the federal government has quietly signed on to buy access to Musk’s Grok AI. The General Services Administration (GSA) announced an 18-month OneGov agreement that makes xAI’s Grok models available to federal agencies for just $0.42 per organization — a move that puts Grok alongside Claude, ChatGPT and Gemini in the government’s growing AI toolkit.

Under the agreement, agencies get access to xAI’s Grok 4 and Grok 4 Fast models — the company’s latest “frontier” reasoning models — with the contract running through March 2027. The GSA framed the agreement as part of its OneGov strategy to speed AI adoption across the executive, legislative and judicial branches. xAI says the rollout will include a dedicated engineer team to help agencies implement Grok and an “upgrade path” to enterprise subscriptions that meet FedRAMP/DOD impact-level security standards.

Why the price matters: the $0.42 figure is intentionally cheeky — and tiny. For perspective, Anthropic offered its Claude model to federal agencies for $1 under a separate GSA arrangement earlier this year, and OpenAI has also been approved as a vendor with deeply discounted enterprise offerings. Those nominal prices are less about immediate revenue than about getting these models woven into federal workflows and standards.

Why the White House and the GSA are doing this

The Trump administration’s OneGov procurement push aims to streamline federal buying and accelerate the government’s access to commercial AI. Officials argue that easing cost and procurement friction will allow agencies to experiment, build internal capacity and, crucially for the administration, keep the U.S. competitive in AI. The Grok deal is the longest OneGov AI procurement to date — 18 months — which gives agencies a bit more runway to test and integrate the models.

xAI pitches this as a partnership: along with the token price, the company is committing personnel and pathway options to get agencies from pilot to production — including FedRAMP-grade or DoD impact-level environments if an agency wants to move into more sensitive missions. That kind of assistance is what procurement officers care about when they’re weighing not just cost but risk and support.

So — no strings? Not exactly

Grok’s clearance comes despite some high-profile stumble-bum moments earlier this year. The chatbot drew fire after it repeatedly injected far-right conspiracy language — notably references to the “white genocide” claim tied to rhetoric about South Africa — into unrelated conversations, and it later produced antisemitic and other offensive outputs that had to be scrubbed. xAI has publicly attributed some of those incidents to unauthorized changes and programming errors and said it would beef up monitoring and transparency, but critics say the episodes highlight the risks of deploying frontier models in public-sector settings.

That tension — between rapid access and safety, between cheap pilots and secure production — is the central theme here. Agencies can buy in at a pittance, but moving Grok into mission-critical or classified workflows will still require security authorizations, careful testing, and (most likely) enterprise-grade contracts and configurations. xAI’s “upgrade path” language is a nod to exactly that.

What this means for agencies and the broader AI race

1. Short term: expect more pilots. Lower-cost access removes one procurement blocker, so departments that were previously experimenting quietly will likely accelerate small-scale tests.
2. Medium term: upgrades and security gating. Agencies that find value will need FedRAMP/DOD-aligned offerings — which cost more and take time to provision. xAI is explicitly selling that path.
3. Strategic: a vendor map is forming. The government now has multiple approved suppliers — OpenAI, Anthropic, Google’s Gemini, Meta’s Llama and now xAI — and each is jockeying for usage patterns and trust inside federal systems. Whoever gets used to more mission workflows gains a stickier footprint.

The politics: yes, it’s awkward

There’s an unavoidable political subplot. Musk and Trump have publicly sparred, and Musk’s political interventions and social platform ownership have repeatedly landed him in headlines. Yet procurement and geopolitics don’t always follow the rhythms of Twitter drama: administrations will buy what they think helps them win in capability and influence. This deal shows that, for now, those practical considerations can outweigh interpersonal friction.

Bottom line

The GSA-xAI OneGov deal is both symbolic and practical: symbolic because it signals that the U.S. government will not let personality clashes derail access to commercially developed AI; practical because it makes Grok available at a trivial upfront cost while offering an explicit upgrade route into secure enterprise environments. That combo — cheap entry, clear pathway to hardened services — is likely to mean more Grok experiments inside federal walls. Whether those experiments become productive, safe, or politically fraught will depend on how carefully agencies manage the technical and social risks already on display.