A U.S. solar equipment manufacturer just cut its quote cycle time by 94 percent. The company did not hire consultants or spend a year on a digital transformation program. It let engineers from Rackspace Technology deploy AI workflows on Palantir’s Foundry platform, and the whole project closed in under two months. On Thursday, Rackspace held that deployment up as proof that its riskiest bet in years is starting to pay off.
A hosting company reinvents itself
Rackspace announced on July 9 that Palantir has named it a preferred partner for regulated and sovereign markets, alongside a definitive agreement establishing an operating framework to deploy Palantir Foundry and AIP in mid-market, regulated and sovereign environments. The San Antonio company, best known for two decades of cloud hosting, is repositioning itself as what it calls the operator of the full enterprise AI stack, from governed private cloud to AI agents running in production.
The logic rests on a real shift in the market. Enterprises in regulated industries, banks, hospitals, energy companies, increasingly want control over their compute, their models and their data, and assurance that proprietary knowledge embedded in that data never leaves their environment. Rackspace is pitching itself as the neutral, model-agnostic layer that lets those companies use whichever AI models they want while keeping policy, identity and data boundaries under their own control.
The Palantir partnership is moving fast
Since the two companies first announced their partnership in February, Rackspace has built up roughly 400 Palantir certifications across its sales, engineering, delivery and operations teams, including a global cohort of Palantir-certified forward deployed engineers serving healthcare, financial services, energy and mid-market customers. The solar manufacturer was the first joint deployment, and the speed of that project, production AI inside a regulated supply chain in under eight weeks, is the number Rackspace wants the market to remember.
Palantir has spent 2026 assembling a network of deployment partners to reach customers its own teams cannot cover, and the preferred-partner designation gives Rackspace a privileged position in exactly the industries where compliance requirements scare off most vendors.
The bet is not free
The same announcement carried a harder message for investors. Rackspace cut its full-year 2026 revenue outlook by 150 million dollars, to a midpoint of 2.5 billion, and trimmed expected adjusted EBITDA by 20 million. The company is deliberately walking away from low-margin public cloud resale and basic hosting to reserve capacity and capital for enterprise AI, and it announced a capital raise to accelerate the transition. Preliminary second-quarter results show revenue between 641 and 649 million dollars and a net loss of up to 91 million.
Chief executive Gajen Kandiah argues the pain is front-loaded. The partnerships Rackspace signed this year with AMD, Palantir, Rubrik, Uniphore and VMware by Broadcom, combined with its existing data centre footprint, are meant to unlock a new revenue stream starting in 2027. The plan calls for 15 megawatts of AI compute capacity by the end of 2027 and 30 megawatts by the end of 2028, built on AMD hardware. Rackspace expects each megawatt to generate 15 to 20 million dollars in annual revenue, which works out to 450 to 600 million dollars a year at full deployment, at adjusted EBITDA margins above 50 percent.
What it signals for everyone else
Rackspace’s pivot is a data point in a larger story: the enterprise AI market is splitting into companies that build models and companies that get them working inside real businesses. Microsoft launched a 2.5 billion dollar deployment business with 6,000 experts earlier this month for the same reason. The money is moving from training AI to installing it.
The open question is execution. Rackspace is spending through losses to chase a market where GPU supply, data centre capacity and customer patience are all constrained. The 2027 numbers will show whether a 25-year-old hosting company can out-deliver the giants in the one segment, regulated industries, where trust still beats scale. For more coverage of enterprise AI, visit Mylistingo.
