A Brutal Guidance Cut Exposes the Cloud Reseller Squeeze

The setup — The broader technology market is catching a bid on the back of fresh frontier models and memory chip debuts, but the infrastructure layer remains unforgiving. While the builders of base compute and top-tier AI models consolidate power, the middlemen who resold cloud capacity are being systematically squeezed out. That tension just caught up to Rackspace.

What's moving

OpenAI unleashes GPT-5.6
OpenAI secured the government greenlight to release GPT-5.6 to the public, rebranding its coding assistant as "ChatGPT Work" while quietly killing off its standalone browser (per The Verge). The launch directly targets the agentic coding market—software that can independently write and execute code over long durations—just as Meta pushes its upgraded Muse Spark model into the same arena (per CNBC). It is a race to lock down enterprise developers before the cost of switching AI agents becomes too high.

The memory giants double down
South Korean memory heavyweight SK Hynix makes its Nasdaq debut today, testing whether a US listing can erase the historic valuation discount attached to its home exchange (per CNBC). At the same time, Micron ($MU) pledged billions more in stateside chipmaking investments, driving a 5% gain as the market bets heavily on a localized memory supply chain to feed data center buildouts (per CNBC).

OpenAI's front-office friction
Beneath the product launches, OpenAI is navigating internal and external turbulence. Fidji Simo, the company's second-in-command for artificial general intelligence development, is stepping down to an advisory role due to chronic illness (per TechCrunch). Simultaneously, news publishers are escalating their copyright lawsuit, accusing the AI firm of hiding datasets that could identify scraped journalism (per TechCrunch).

Featured: Rackspace Technology, Inc. ($RXT)

The move
Rackspace lost a third of its value in a single session, falling 33.59% to close at $4.37. The stock traded on zero relative volume momentum, signaling a complete institutional reset that erased months of stabilization in the share price.

What drove it
The catalyst was a severe fiscal 2026 guidance cut paired with a dilutive $250 million stock offering (per Yahoo Finance). Rackspace slashed its revenue target by $150 million to a midpoint of $2.5 billion, and lowered its adjusted EBITDA forecast to a range of $285 million to $295 million. The headline numbers were bleak enough to completely bury a newly expanded AI partnership with Palantir ($PLTR). The real damage came from management's admission that they are intentionally walking away from low-margin public cloud resale operations, a segment that took a $125 million hit in the guidance revision.

The bigger picture
Rackspace is caught in the most dangerous place in modern enterprise technology: the middle. For years, companies built a business model on wrapping support services around computing capacity from the major public clouds. Now, hyperscalers—the dominant cloud operators providing the underlying servers—are increasingly bypassing these middlemen to sell directly to end users.

Rackspace is trying to pivot into an enterprise AI infrastructure provider, targeting 30 megawatts of data center capacity by 2028. But building out hard assets requires intense capital outlays precisely when cash flow from the legacy cloud hosting business is drying up. The transition from a services layer to a physical AI hardware player is expensive, and as Revenue Sharing with Cloud Partners Reshapes AI Infrastructure demonstrated earlier this week, the major cloud operators are retaining the lion's share of the economics.

Across the tape

The Nasdaq 100 outpaced the broader market, gaining 1.66% as the 10-year Treasury yield drifted lower to 4.54%. Meta ($META) saw its shares rebound as the market digested news that its new modular, custom AI chips will enter production in September, easing fears about unchecked third-party compute spending (per TechCrunch).

Elsewhere, Palo Alto Networks ($PANW) CEO Nikesh Arora warned that AI token pricing needs to fall 90% for businesses to adopt the technology at scale, putting pressure on model builders to deliver better compute efficiency (per CNBC). On the power front, clean energy continues to dominate new electrical capacity additions, accounting for roughly 90% of grid expansion as data center demands strain existing infrastructure (per CNBC).

What to watch

• SK Hynix's trading volume: Watch how the market prices the South Korean memory giant in its first few days of US trading, which will set a new benchmark for pure-play memory multiples.
• Rackspace's offering price: Look for the final pricing of Rackspace's $250 million stock offering. A steep discount will signal how little appetite Wall Street has for funding legacy cloud pivots.
• Palo Alto's margins: With Arora publicly calling out token costs, watch for margin commentary in upcoming cybersecurity earnings prints to see if AI add-ons are eating into free cash flow.
• OpenAI's enterprise adoption: Track the early uptake of "ChatGPT Work," which will test whether the new GPT-5.6 model's claimed 54% token efficiency improvement translates into faster corporate deployment.