Zhipu reportedly reached $1 billion in annual recurring revenue in July, roughly four times a March estimate, but the unconfirmed run rate is not annual sales and still sits far ahead of recognized cloud revenue while margins remain thin.
A new U.S. rule puts the UAE in a favored export-control tier and lets a named group receive certain AI chips and other controlled technology without individual licenses, reducing recurring reviews while leaving end-use rules, entity limits and a 270-day clock for G42 and Core42 in place.
The United States now lets a named group of chip buyers and technology operators in the United Arab Emirates receive covered items without individual export licenses. The change is commercially important, but narrower than a blanket opening: the approved list, existing end-use rules and a temporary authorization for two Emirati companies still define who can receive what.
Effective July 10, the Bureau of Industry and Security moved the UAE into Country Group A:5 and removed it from the D:3 and D:4 groups associated with chemical, biological and missile-technology concerns. The final rule, published July 14, expands eligibility for certain Commerce-controlled military equipment, commercial spacecraft and dual-use technology used in industries including oil and gas, desalination and civil nuclear power. Removal from D:3 and D:4 also eliminates restrictions on support for UAE unmanned-aircraft programs, BIS said in its announcement.
A:5 includes NATO members and allies such as Japan and South Korea. The UAE is the first Arab country given the designation. It is also the only A:5 member that does not participate in the multilateral export-control regimes associated with the group, and neither Israel nor Saudi Arabia is a member, according to a contemporaneous account.
The label does not give every UAE buyer the same privileges. The new supplement bars use of the Strategic Trade Authorization, or STA, exception when an ultimate consignee or end user in the UAE is not approved for it. Part 744 end-use and end-user license requirements still apply, and shipments under STA still require export-system filings and recordkeeping.
| Recipient in the UAE | Authorization without an individual license | Main boundary |
|---|---|---|
| UAE government agencies, including the Ministry of Defense | Certain advanced-computing items and other items eligible for STA | Excludes state-owned companies, contractors and grantees |
| G42 and Core42 | Certain advanced-computing items | No broader STA authorization; expires automatically after 270 days absent another BIS notice |
| Amazon, Apple, Google, Meta, Microsoft, OpenAI, Oracle and xAI, plus subsidiaries | Certain advanced-computing items and other items eligible for STA | STA conditions and other end-use controls remain |
| MGX | None under the new exception | Must apply for licenses; BIS says it intends favorable review for semiconductor and server applications |
Other entities can ask to be added through an advisory-opinion process. The rule gives the Commerce secretary 30 days to decide, in consultation with the State Department and the White House national-security adviser, and BIS five days after a decision to notify the applicant.
That distribution matters more than the A:5 label alone. G42 and Core42 receive temporary, item-specific relief. The eight U.S.-headquartered companies receive both the advanced-computing authorization and the broader STA benefit, with no comparable expiration stated in the rule.
The new default should reduce recurring regulatory reviews for approved recipients, but the government’s own paperwork estimate is modest: about 50 fewer license applications a year, 25 fewer burden hours and $950 in savings. BIS also expects the workload associated with license exceptions to increase.
Nor was licensing an absolute barrier before the change. Microsoft said in its account of its UAE investment that earlier approvals enabled it to accumulate capacity equivalent to 21,500 Nvidia A100 processors, using a mix of A100, H100 and H200 chips. Licenses approved in September 2025 allowed the company to ship another 60,400 A100-equivalent chips using GB300 processors. Microsoft said both rounds depended on cybersecurity, national-security and technology safeguards.
The rule therefore replaces individual review for covered transactions with listed approvals and exception conditions; it does not change the physical economics of deployment. It contains no chip allocation, construction financing, power commitment or completion schedule.
Projects that could use the authorization are nevertheless large. Microsoft said it would spend $15.2 billion in the UAE from 2023 through 2029. That total includes a $1.5 billion equity investment in G42 and more than $4.6 billion of AI and cloud data-center capital spending through 2025. Of the $7.9 billion Microsoft planned to spend from 2026 through 2029, more than $5.5 billion was earmarked for AI and cloud infrastructure. Those are Microsoft’s investments across the country, not a financing commitment for the separate 5-gigawatt campus.
OpenAI said in its project announcement that Stargate UAE would be a 1-gigawatt Abu Dhabi cluster involving G42, Oracle, Nvidia, Cisco and SoftBank, with 200 megawatts expected to go live in 2026. It also described UAE investment in U.S. Stargate infrastructure. The earlier bilateral framework placed that 1-gigawatt data center inside a planned 5-gigawatt UAE-U.S. technology cluster and called for robust U.S. security standards.
Those statements establish scale and intent, not delivery. The archived material does not show how much of the first 200 megawatts is operating, how many chips will be installed, or whether the wider campus has secured the infrastructure and capital required for completion.
The administration linked the upgrade to the UAE’s status as a Major Defense Partner and its support for U.S. national-security interests, including Operation Epic Fury. The UAE had carried out airstrikes against Iran, intercepted missiles and helped keep oil moving through the Strait of Hormuz, according to an account of the decision. A separate report on the announcement described the defense support, including during the Iran war, as the stated basis for the upgrade.
The UAE calls the decision recognition of a relationship that spans security, technology, trade and investment. Ambassador Yousef Al Otaiba said in a statement that the cooperation had developed across six U.S. presidential administrations. Emirati officials have also presented A:5 status as validation of the country’s export-control and compliance framework.
The unresolved question is whether compliance after approval can manage the risks that previously drove tighter restrictions. U.S. officials had limited advanced-chip sales because of concern that the UAE’s relationship with China could expose sensitive technology. U.S. intelligence agencies obtained information in 2022 indicating that G42 had provided technology used by China’s People’s Liberation Army to extend the range of some missile systems, according to reporting on the policy; G42 strongly denied the allegation. The UAE and U.S. officials said stringent safeguards protect the technology.
Microsoft offers one example of those controls. The company said its $1.5 billion G42 investment was accompanied by an intergovernmental assurance agreement covering cybersecurity, physical security, export controls, technology transfer, data protection and customer screening. That is a company description of its own compliance framework, not independent evidence that it has prevented every prohibited transfer.
Outside experts cited in the same policy report split on the competitive trade-off. Former U.S. official Chris McGuire said the decision could put large data centers in the UAE and expose them to Chinese access. Ryan Fedasiuk of the American Enterprise Institute said Washington should have demanded a stronger firewall against Huawei, but argued that American chips running American workloads abroad could improve U.S. competitiveness as domestic resistance to data-center construction grows.
Political influence is a separate challenge to the decision’s credibility. Senator Elizabeth Warren alleged in a committee statement that the UAE royal associated with G42 and MGX secretly bought 49% of the Trump family’s World Liberty Financial and that President Donald Trump received a $263 million windfall related to the deal. Reporting identified the buyer as Sheikh Tahnoon bin Zayed Al Nahyan, who chairs G42, and put the purchase price at $500 million. Warren called for Commerce Secretary Howard Lutnick and Under Secretary Jeffrey Kessler to testify. Emirati and U.S. officials have denied that UAE investments influenced access to advanced chips.
The first explicit decision point is the automatic expiration of G42 and Core42’s authorization 270 days after the rule was filed, unless BIS issues a subsequent notice. The rule does not describe that step as an automatic renewal or state the evidence BIS will require before extending the benefit.
That makes the next record of enforcement more revealing than another capacity announcement. The central questions are whether BIS discloses the chip volumes and approved end uses, whether safeguards detect prohibited customers or technology transfers, and what grounds the agency gives for extending or ending the two Emirati companies’ temporary access.
Construction evidence matters too. The announced 200-megawatt phase, the larger 1-gigawatt cluster and the planned 5-gigawatt campus are three different scopes. Until operators report what is financed, energized and in service—and until the governments show how matching U.S. investment commitments are being measured—the rule should be read as a regulatory authorization for selected recipients, not proof that the promised infrastructure exists.
Anthropic’s 20-year lease gives TeraWulf a customer for 401 megawatts of future AI infrastructure, but rent starts only after delivery and the proposed utility deal passes power and grid costs to TeraWulf, leaving financing and project margin unresolved.
OpenAI has put conversations and Projects back in its redesigned ChatGPT desktop app and enabled cloud Work threads to move across devices, correcting the launch's biggest usability failures without merging local Work or Codex histories.
Twenty-nine countries signed an agreement creating WAICO as an independent intergovernmental organization, while China paired the launch with capacity-building offers that are not yet confirmed as WAICO programs.
Meta reportedly plans to put departing AWS compute executive Dave Brown to work on its data-center buildout, adding hyperscale operating experience while leaving any customer-facing cloud business conditional and undefined.
Moonshot AI has made Kimi K3 available through its apps and API, pairing a 2.8-trillion-parameter architecture with early frontier-level results, but the model's open-weight claim cannot be tested until its weights and technical report arrive.
CIA Director John Ratcliffe said US intelligence is consistent with an estimate that Russian recruits last 20 to 30 minutes on Ukraine’s battlefield, but the public trail leads to an unsourced claim about assault troops and does not establish a representative average.
China has paired a five-year AI training offer for developing countries with cooperation centers, a weather-warning rollout and a new 29-country organization. The package gives Beijing a platform for influence, but no budget, selection rules or delivery timetable has been published.
Nebius has arranged its first senior secured facility against an operating GPU deployment and one customer’s cash flows. The deal adds project-level debt to its expansion toolkit, but an unnamed customer and a larger rival financing limit what it proves about the rest of Nebius’s backlog.
Huawei publicly displayed a 16-cabinet Atlas 950 configuration rated at 1 EFLOPS in FP8, providing tangible evidence of its system-scale AI strategy while leaving price, power use and sustained workload performance undisclosed ahead of the full system's planned fourth-quarter release.
UK testing places leading open-weight models four to seven months behind selected closed-model cyber results, yet longer attack chains, U.S. benchmarks and mixed cost comparisons show why that interval is a warning signal rather than a universal capability clock.
Moonshot AI's Kimi K3 added to a global technology selloff with near-frontier performance, but unreleased weights, mixed cost comparisons and a recommended 64-accelerator deployment leave its effect on chip demand unresolved.
Two binding EU decisions require Google to give rival AI services comparable access to 11 Android features and offer eligible search competitors a restricted, anonymized dataset, but phased deadlines, certification, pricing and privacy safeguards leave the competitive effect unproven.
Databricks has signed a term sheet for a Coatue-led financing at a $188 billion valuation, while unidentified sources put the round at $3 billion. The proposed capital would deepen its push into AI governance, data agents and operational databases, but the transaction remains open and the company supplied no new operating figures.
OpenAI says a handful of GPT-5.6 Sol file-deletion reports most commonly involved Full Access without sandboxing or Auto-review. Its own evaluations show a more complicated risk picture, and the company has not yet published an incident rate or evidence that its promised safeguards stop the failure.
Netflix says generative AI workflows were used on roughly 300 titles in 2026, but its only quantified example tied to that disclosure covers 17 minutes and does not establish how much finished material, spending or labor changed across the slate.
A rogue browser extension can trigger Claude for Chrome’s fixed Gmail, Docs, Calendar and business workflows; the attack is constrained by default approvals but can run silently for users who enabled unattended action.
Gemini 3.5 Pro missed its expected June rollout. An anonymous-source account says a late-June data change fell short of Google's coding goals, but Google has confirmed only partner testing—not the reported cause, a new date, or public results and pricing.
Gold Eagle has begun collecting and prioritizing AI-discovered software vulnerabilities, but the voluntary federal clearinghouse has not disclosed results, operating rules or the resources that would turn findings into deployed fixes.
Google is bringing Gemini Omni editing and a reusable face-and-voice avatar to Vids, but the sharper distinction is account-level identity across Vids and Gemini rather than a new category of AI video; Vids already offered Veo generation and customizable avatars, while specialist rivals already sell digital presenters.