TheChek News Flow

DAX:

MDAX:

SDAX:

WASHINGTON/RIYADH, July 22 (Reuters) - Iranian drone attacks on CIA facilities in the Gulf have prompted U.S. intelligence analysts to investigate whether Russia assisted by providing targeting information or ​advanced drone technology, said four people familiar with U.S. intelligence.

These people, who spoke on condition of anonymity to discuss national security matters, said U.S. intelligence officials have ‌not yet reached firm conclusions about the possible Russian involvement in the attacks on CIA facilities. But they cited the strikes' effectiveness and apparent precision, as well as Russia's broader technical support for Iran, as possible evidence.

Reuters and other media have reported — and U.S. officials have acknowledged — that Russia has provided targeting and other support for Iran in its strikes on U.S. targets in the Gulf region generally, but the intelligence community's investigation of ​possible Russian involvement in supporting the CIA facility attacks has not been previously disclosed.

At least two CIA sites were struck in March, Reuters and other outlets have reported. One ​of the facilities was the CIA station in Saudi Arabia, which is located in the U.S. embassy in Riyadh, and a separate site was ⁠located in eastern Iraq. Some of the sources said additional CIA sites had been hit, but they did not disclose details.

One internal memo from a Western intelligence ​agency, described by an official in the region with access to the document, concluded Russia has likely played a role in targeting regional CIA facilities. The official discussed the memo on condition ​that its precise authorship not be disclosed.

Two Western officials in the Gulf who were briefed on intelligence reports said analysts thought the Saudi attack involved a pair of Russian-enhanced versions of Iran's Shahed drones. One of the aircraft blew a hole through a vulnerable part of the embassy's exterior and the second flew through the opening and detonated, they said. No deaths or injuries were reported.

While Russian assistance for Iran is longstanding, given the ​countries' close ties, specific targeting of sensitive CIA sites would indicate Moscow is willing to go further in disrupting U.S. operations as Washington struggles to bring the Iran war to an end.

Concerns ​that Russia could be providing targeting data to Iran intensified this weekend after strikes on a U.S. air base in Jordan, where Iran struck army barracks with ballistic missiles, killing two soldiers.

The White House referred a request ‌for comment ⁠to the CIA, which declined to provide one.

Iran's mission to the UN, the Russian Defense Ministry and the Saudi government did not respond to requests for comment.

The two Western officials in the region said Russia has helped Iran improve its Shahed drones' ability to reach targets, and said analysts suspected that Iran used these versions in its strike in Riyadh.

Russia, said a separate source, has helped Tehran improve the accuracy of its Shahed-136s by supplying a satellite navigation system — the Kometa-M — that experts say is far more accurate and harder to jam than the one produced by ​Iran.

The Wall Street Journal first reported in March ​that Russia has provided the Kometa-M to ⁠Iran. The Kremlin denied the report, calling it "fake news."

CIA facilities overseas include the agency's main offices in individual countries that traditionally are housed in embassies and are called stations. Additionally, the CIA maintains offices, safe houses, logistics hubs and sites involved in surveillance or other operations.

The locations of ​those sites are closely guarded secrets.

The sources declined to disclose the exact number or locations of CIA facilities hit by Iranian drones. ​One source put the number ⁠at "more than one and fewer than a dozen." A second source said several facilities were hit.

U.S. intelligence analysts have been looking into whether Iran relied on Russian targeting data for the embassy attack, said the sources.

Doubts remain, however. Some of the sources warned that Iran could have been aiming for the U.S. embassy generally and by chance struck the CIA station.

Some analysts have concluded that few ⁠nations beyond Russia ​would have the means to acquire and the interest in weaponizing such sensitive targeting intelligence against the U.S., according ​to the sources.

Daniel Hoffman, a former CIA station chief and undercover operations officer, said it would not be surprising for Moscow to help Tehran target CIA facilities given their close strategic partnership.

"They share an interest in trying to reduce ​or eliminate U.S. influence altogether in their self-designated spheres of influence," he said.

Cloud and software incumbents want businesses to think twice before using Anthropic, OpenAI and other closed-source AI providers. Executives such as Palantir CEO Alex Karp have argued, for instance, that these AI firms could use their customers’ data to compete with them eventually.

These potential fears could also help new AI startups catering to businesses.

One such startup is Noeri, a new AI lab founded by two Stanford computer science professors to help businesses customize AI models for their specific needs so they don’t need to buy models from Anthropic and major providers. Menlo Ventures, a major Anthropic backer, is in talks to lead a $100 million funding of the startup, according to a person involved in the discussions.

Other venture firms including Andreessen Horowitz, New Enterprise Associates, Madrona as well as Nvidia are also in talks to invest as well, the person said.

Their interest in the startup is a sign that investors think businesses will be more wary of working with the most advanced AI providers as those providers—namely, Anthropic—keep launching features and applications targeting a variety of professional workers. Anthropic, for instance, blindsided some business partners that had used Claude to power apps for designers or lawyers when it later launched its own, competing apps or features for people in those fields.

The funding round will likely have multiple tranches at different valuations, a common occurrence among hot AI rounds these days, but the final valuation could range anywhere from $600 million to $1 billion, the person said. (Different valuations happen when there’s too much demand for a funding round, leading some VCs to offer to invest at higher valuations to sweeten the deal for the startup.)

Noeri was founded earlier this year by Yejin Choi and Carlos Guestrin, who are well-known for their work on test-time training and reinforcement learning via self-distillation, terms that refer to helping a model learn and improve on certain tasks after it’s been trained. (Guestrin’s work on self-distillation in particular was used by Cursor in training its latest coding models.)

Noeri is far from the first company to try and help businesses tweak models for specific tasks (otherwise known as finetuning). Thinking Machines Lab’s Tinker product, for instance, provides finetuning software that developers can access through an application programming interface, while Applied Compute takes a more hands-on approach by embedding forward deployed engineers with its customers.

Noeri hasn’t decided on its strategy or business model yet, the person with knowledge of the discussions said. It’s possible that the company will do the finetuning work itself for customers and provide those customers with the finished model at the end of the process. It’s also possible that it’ll take a more hands-off approach, a la Tinker.

The Takeaway

Elon Musk’s AI company, SpaceXAI, is laying the groundwork for at least one new large-scale data center in Texas, three people with knowledge of the effort said. Such a move could substantially expand SpaceX’s AI computing capacity outside its existing data center hub in Memphis, Tenn.

An expansion could position SpaceXAI to be a major cloud provider, as its existing AI business has struggled to gain traction and the company has been striking a series of deals to rent capacity to outside companies.

SpaceXAI in recent months has explored several potential data center sites in Texas, two of the people said. It has discussed various approaches to the expansion, the two people said, including setting up multiple data centers both by developing from scratch and by retrofitting an existing warehouse, similar to how SpaceXAI repurposed a former manufacturing facility for its first Memphis data center.

The goal is an expansion in Texas on a similar or greater scale to the company’s existing facilities in the Memphis area, one of the people said. SpaceX says it has around 1 gigawatt of compute capacity between two operating data centers, with hundreds of thousands of Nvidia graphics processing units installed.

SpaceXAI has already sent some of its existing data center staffers to Texas in preparation for an expansion there, two people said. The company was hiring for a data center development lead last month in Austin and Bastrop, Texas, to oversee data center installations and provide support with matters including “identifying prospective properties of interest, negotiating lease and own acquisitions,” according to the listing. And earlier this month, SpaceX posted job openings for temporary roles to support its “data center construction, expansion, and operations” in locations including Conroe, Texas.

SpaceX has touted its ambitions to one day launch and operate a big network of orbital data centers, but its AI infrastructure expansion is for now very much earthbound. The company has said it may continue to expand terrestrial data centers beyond its existing facilities, but it has disclosed little publicly about where exactly it plans to add more AI computing capacity.

Further expansion could extend its already heavy spending on AI infrastructure. SpaceX’s AI-related capital expenditures jumped to $7.7 billion for the first quarter, roughly triple the number for the same period a year earlier. Ahead of its June initial public offering, Morgan Stanley analysts projected SpaceX would burn $120 billion in 2026 and 2027 combined, due largely to AI-related capital expenditures.

Supermicro CEO Charles Liang said on X in June that the company was helping SpaceXAI build “another new Gigawatt AI datacenter” within a year, though he did not specify where the project would be located. The AI server maker has previously touted its role supplying SpaceXAI with Nvidia-based servers and cooling equipment.

The timeline for how quickly SpaceX could get new data centers up and running is unclear. Representatives for Conroe, Texas, did not respond to multiple requests for comment. SpaceX did not respond to a request for comment.

SpaceX had touted its ability to rapidly add data center capacity as a key advantage for its AI business in the run-up to its IPO. But its own Grok offerings have struggled to gain traction with business customers, prompting Musk to overhaul the company’s AI product strategy in recent months.

In June, SpaceXAI announced deals to lease large parts of its compute capacity to other companies including Anthropic and Google. This fresh source of AI revenue comes on the heels of SpaceX burning billions of dollars in cash in the first quarter, due largely to heavy AI capital expenditures.

Still, Musk has said SpaceXAI can take back capacity if needed. SpaceX has also agreed to a $60 billion deal to acquire AI startup Cursor, which is aiming to broaden beyond its coding roots and develop leading general-purpose models by the end of this year.

SpaceXAI has dramatically expanded its data center workforce over the past year, adding more than 40 new hires to its Colossus team, according to LinkedIn data. The AI unit has also brought in SpaceX employees to assist with its data center operations following the companies’ February merger, one person familiar with the work said.

SpaceXAI has been exploring a potential Texas data center expansion for several months, one person said. The company has faced a series of reorganizations and leadership churn over the past year. Notably, xAI co-founder Ross Nordeen left the company in March. He had overseen the company’s data center build-out and frequently updated Musk on its progress, two people said.

SpaceXAI President Michael Nicolls appointed new leadership in early April, telling staff in a memo that Jake Palmer would lead physical infrastructure and Daniel Dueri, director of software engineering at SpaceX, would lead compute infrastructure.

The AI business has so far relied heavily on natural gas turbines to supplement available grid power and power its existing data centers. It has installed dozens of mobile turbines in Tennessee and Mississippi, and it is developing a permanent gas-fired power plant in Mississippi through a joint venture with Houston-based Solaris Energy Infrastructure.

The company also has started work on further expansion in the Memphis area, including a third data center. Late last year, SpaceXAI purchased an 810,000-square-foot warehouse near its other data centers just across the state line in Mississippi.

But SpaceXAI has faced backlash in Memphis from local residents, politicians and environmental groups over its use of gas turbines as well as other issues including water use. Amid cost cutting earlier this year, the AI unit delayed work on a $80 million facility intended to recycle wastewater from the data centers and other industrial sites.

SpaceXAI has also sent staffers to Florida as well as Texas in recent weeks, one of the people said. The company posted data center roles for Jacksonville, Fla., in addition to Texas in recent weeks.

A spokesperson for the city of Jacksonville said last week that the city was not aware of any plans for a data center and had not had any communications with SpaceXAI. The spokesperson did not return requests for comment this week.

Musk and his companies have existing ties to both areas, with SpaceX and Tesla jointly building a semiconductor manufacturing facility in Grimes County, near Conroe. And Musk acquired Jacksonville-headquartered power generation provider APR Energy in May, a Federal Trade Commission filing showed.

APR Energy operates in several states where SpaceXAI has set up data centers, including Tennessee, Mississippi and Texas, according to business records with the states. Musk’s AI startup has purchased turbines from APR Energy in the past, one of the people said.

But its strong financial position sets it up to pursue even more aggressive bets. In recent months, companies have grown more focused on reining in their spending on AI technology. Stripe also has relationships that tie into the fast-growing practice of using intermediaries to track and route AI inference requests to lower-cost model providers.

Popular model routing startup OpenRouter, valued at $1.3 billion earlier this year, uses Stripe for payments processing, as well as invoicing, tax and other services, the companies said early this year. OpenRouter has been attracting acquisition interest at billions of dollars from at least one larger tech firm, a far higher price tag than in its most recent funding round, The Information reported last week.

OpenRouter could be a good fit for Stripe: OpenRouter collects money from AI users, charging a fee of up to 5.5% to the AI user on top of what users pay model providers.

Stripe in March also added an AI Gateway service, which allows developers to access different models and track usage so they can bill their own customers. Stripe currently doesn’t charge markup fees for the AI service, and it is giving away some AI tokens to users as incentives, according to tweets by Stripe Product Manager Miles Matthias.

Other payments firms, such as corporate card and expense startup Ramp, are also launching AI gateway features.

At the same time, Stripe is positioning its crypto-focused payments business to benefit from the AI boom. Stripe last year acquired Bridge, a startup that helps businesses convert funds into stablecoins, in a $1.1 billion deal. Bridge issues stablecoins and processes stablecoin payments. Stripe has also backed Tempo last year, a blockchain designed for stablecoin payments, and more recently became a founding member of the consortium behind new stablecoin Open USD, alongside Visa, Mastercard, Coinbase and others.

Stripe has promoted stablecoins as a way for AI agents to make small payments for services such as model access and computing, arguing stablecoins are better suited for real-time micropayments. That would include machine-to-machine payments, where software systems pay each other automatically. Stripe is working on enabling those payments for its AI gateway feature, Matthias said on X.

Stripe declined to comment for this story.

A PayPal deal, meanwhile, would dramatically reshape the broader payments industry, since PayPal owns Venmo and the behind-the-scenes Braintree processing business in addition to its PayPal-branded digital wallets. It could also accelerate Stripe’s long-running efforts to boost margins as it faces pressure from some of its bigger customers.

While Stripe got its start processing payments for a generation of startups including Lyft, Shopify and DoorDash, some of those early Silicon Valley customers are now mature businesses, giving them more leverage over Stripe. Some have pushed to negotiate lower prices, and many have added processors such as Braintree, Adyen and Worldpay. That includes e-commerce giant Shopify, which added PayPal as a back-end processor two years ago.

Any financial outlay on any PayPal deal would likely be much smaller than the $53 billion overall offer. With a private equity partner, any deal would likely involve some borrowed money, and Stripe is reportedly interested primarily in certain parts of PayPal such as Venmo.

One recent focus for Stripe has been pushing to shift more of its payments volume away from credit card–based transactions toward consumer payments made via direct withdrawals from bank accounts, a person familiar with the effort said. That would allow Stripe to avoid paying some credit card network and interchange fees, which are a major cost in its payments business.

Stripe offers Financial Connections, a product that competes with Plaid by letting customers securely link their bank accounts and share account information with businesses. Stripe also has its own digital wallet, Stripe Link, which stores credit cards but also allows consumers to connect directly with their banks to pay online.

But Stripe isn’t well known as a consumer brand compared to PayPal wallets and Venmo, which have millions of customer accounts that are already linked to bank accounts. That makes PayPal’s consumer businesses particularly attractive to Stripe.

To be sure, PayPal hasn’t responded publicly to the offer yet, suggesting it could seek a higher price or keep trying for a turnaround under its new CEO. PayPal is set to report second-quarter earnings next Tuesday.

Alphabet investors have very publicly worried that the company’s massive AI spending isn’t worth the money. With the company’s latest earnings report, those investors should be able to relax a little.

The takeaway: Google’s cloud business — driven largely by enterprise AI adoption — is booming. The search giant saw Google Cloud revenue spike 82% from where it was this time last year, climbing to $24.8 billion. That’s well above last quarter’s generous year-over-year growth, which showed a revenue jump of 63% to $20 billion — and it handily beats what Wall Street analysts expected for this quarter’s growth (the expectation was $22.46 billion).

Those cloud gains were driven largely by enterprise AI solutions and enterprise AI infrastructure adoption, the company said, while also noting that its backlog of cloud contracting work — that is, work that it hasn’t yet converted into revenue — had climbed to $514 billion.

The company’s profit hit $112.1 billion, which is a massive jump from this time last year, when the company reported $28.1 billion in profit, the company’s earnings report shows. Meanwhile, Alphabet’s overall revenue grew 24% year-over-year during the past quarter to $119.8 billion. The company also saw Google Services revenue jump 15% to $94.5 billion.

“Our AI investments are redefining what’s possible across every part of our business,” said Google CEO Sundar Pichai during Wednesday’s earnings call. “We have exciting momentum across the board.”

More people are also adopting Gemini, Google’s AI chatbot, as the app currently enjoys 950 million monthly active users, the company said. In Q4 of 2025, Google reported that the app had 750 million users.

It’s worth noting that spiking revenue isn’t unusual for Google. This marks the company’s 12th consecutive quarter of double-digit revenue growth. But even by that standard, this quarter represents a particularly bountiful period for the tech giant.

Alphabet’s spending is still hefty, with its capital expenditures — the money it spends building data centers, buying chips, and expanding infrastructure — estimated to be between $180 billion and $190 billion for the year — a fact not lost on analysts during Wednesday’s earnings call. Several pressed Pichai on when, and how much, those investments will pay off.

“I think our compute capacity investments in ’27,” he said. “We are seeing strong demand indicators, including long-term deals,” he continued. “I think, if anything, the dynamics look healthier than where we were about a year ago, so that’s what gives us the confidence to undertake those investments,” he said.