• Amazon is prepping for customer questions following its $50 billion investment in OpenAI.
• Amazon employees are instructed to stress its continued commitment to Anthropic and Nova models.
• It also warns against saying customers can “access” OpenAI models through AWS.

When Amazon announced a $50 billion investment and sweeping cloud partnership with OpenAI recently, a question surfaced inside the cloud giant: What does this mean for Anthropic?

Amazon is one of Anthropic’s biggest investors, and has a deep cloud partnership with the startup. OpenAI and Anthropic are arch enemies, so there are potential tensions and conflicts from Amazon working so closely with both AI labs.

The tech giant addressed some of this in internal talking points it prepared recently for sales and marketing teams at Amazon. Business Insider reviewed a copy of the guidelines.

The document is the latest example of how intertwined relationships have become at the leading edge of the generative AI field. This technology requires a massively expensive rewiring of the cloud industry and the tech sector in general.

That’s led to tech giants investing heavily in many of the top AI startups. For instance, Anthropic has taken billions of dollars from Google and Amazon, which are themselves arch rivals in cloud computing and online product search.

“Within the guardrails”

The Anthropic issue is just one of several delicate topics covered in the memo. The guidance lays out approved language, prohibited phrasing, and prepared responses to questions ranging from competitive dynamics to accusations that the OpenAI arrangement amounts to a circular financing.

Taken together, the materials reveal how deliberately Amazon is shaping the narrative around one of the most consequential AI alliances in the industry.

“It is very important that all our marketing stays within the guardrails,” the memo said. An Amazon spokesperson declined to comment.

Reassurance for AWS customers

In its guidance, Amazon instructed employees to reassure AWS customers that it maintains “strong relationships” with Anthropic as well as other AI model providers including Meta, Mistral, and Cohere.

“We will continue to work closely with all model providers and only expect these partnerships to strengthen over time as customer demand for multiple models increases,” the document stated.

Not “OpenAI on AWS”

As part of Amazon’s new deal with OpenAI, the companies created an AI system architecture called Stateful Runtime Environment, or SRE. The new service is powered by OpenAI models and available on Amazon Bedrock, the cloud giant’s platform for customers to access various models.

Amazon is tightly controlling how employees describe the SRE offering.

AWS employees may say the SRE “is powered by OpenAI models,” “is enabled by OpenAI models,” or “integrates with OpenAI models,” according to the talking points in the document reviewed by Business Insider.

However, AWS staff are explicitly told not to say SRE “enables access to OpenAI models” or allows customers to “call OpenAI models.” The document warns against describing the SRE as a “passthrough” to GPT models or suggesting that OpenAI’s frontier models are generally available on AWS.

AWS employees are also told not to imply that OpenAI is “offering” the SRE. Instead, the companies are “jointly collaborating to offer” it.

The distinction is deliberate. OpenAI models will underpin the SRE, but customers can’t directly call them through existing Bedrock APIs. (Application Programming Interfaces, or APIs, are a common way for applications to access and share data).

That positioning by Amazon separates its new deal from Microsoft’s arrangement of hosting OpenAI models on its Azure cloud service. It also reinforces that AWS is not merely reselling OpenAI models, but embedding its models inside a specific infrastructure service.

Many operational details still remain undisclosed. Pricing, technical limits, and regional availability are all listed internally as “stay tuned.”

Is this a circular deal?

The internal memo also addresses the question of whether Amazon’s $50 billion investment in OpenAI amounts to a circular deal. Such concerns have become common in the AI industry where tech giants invest in startups, only to see a lot of that money come back in the form of cloud spending.

In Amazon’s case, OpenAI is expanding its existing AWS cloud agreement by $100 billion over 8 years and committing to use 2 gigawatts of AWS Trainium chips as part of last month’s deal.

AWS employees are instructed to push back on any circular financing claims. The document argued that companies frequently invest in and do business with one another, particularly in capital-intensive industries. It says Amazon’s reasons for investing in OpenAI are distinct from OpenAI’s reasons for using AWS infrastructure.

Giving up on Nova?

The talking points further prepare teams to respond to concerns that the arrangement could sideline Amazon’s own Nova AI models or its Quick agentic AI application offering. As part of the deal, Amazon will become the exclusive provider of OpenAI’s Frontier service, which has enterprise technology features that are similar to Quick.

The guidance emphasized Amazon’s continued commitment to Nova and Quick, and the importance of providing choice to clients. Customers use multiple AI models even within the same application, the guidance stated.

The talking points also address chip supply concerns.

Given OpenAI’s massive infrastructure needs, AWS anticipates customers asking whether Trainium capacity will be constrained. Amazon has previously said AWS growth would have been stronger if not for tight capacity issues.

The guidance, however, reassures teams that many customers will still have access to Trainium for their own AI workloads.

Have a tip? Contact this reporter via email at ekim@businessinsider.com or Signal, Telegram, or WhatsApp at 650-942-3061. Use a personal email address, a nonwork WiFi network, and a nonwork device; here’s our guide to sharing information securely.

• Citigroup is among US firms warning staff to evacuate or stay in safe locations in the Middle East.
• The firm has turned to a “fully remote model” as the crisis in the Middle East continues.
• On Wednesday, reports suggested Iran could strike regional banks and financial targets.

Citigroup evacuated three of its buildings in the United Arab Emirates on Wednesday, as the crisis in the Middle East is prompting US banks to issue updated security guidance to employees.

Representatives for Citi confirmed to Business Insider on Wednesday afternoon that the company implemented a “fully remote model” as it assesses the situation in the UAE, which is separated from Iran by only the narrow Strait of Hormuz.

The bank had previously instructed employees to work from home.”The vast majority of our people in the UAE have been working remotely, and we have now moved to a fully remote model for all UAE-based colleagues,” representatives said. “We are continuing to serve our clients without interruption. The decision to evacuate three of our buildings in the UAE was responsive to information we received and is consistent with our commitment to prioritize the safety of our colleagues. All colleagues are accounted for and are safe,” the representatives added.

The move comes after Iranian military leaders warned on Wednesday that they would attack banks and financial institutions in the region, according to reports in the Associated Press, Reuters, and Al Jazeera, following US strikes on the country that began in late February and resulted in the death of longtime Supreme Leader Ali Khamenei.

Some of the UAE’s major cities, like Dubai and Abu Dhabi, are key regional industrial and financial centers.

Other financial firms are monitoring the situation closely as well. A person with close knowledge of the policies at the Wall Street giant Goldman Sachs said the firm’s main priority was employee safety and that it had instructed all its teams, regionally, to work from home and adhere to local guidance.

• Most of Iran’s best air defense systems are no longer factors, Gen. Caine said Tuesday.
• US and Israeli forces are achieving air superiority and conducting deeper strikes in Iran.
• Both the US and Israel have suppressed and targeted Iranian air defenses.

Most of Iran’s best air defenses are no longer factors in mission plans for American air operations, giving US and Israeli air forces space for strikes deeper in the country, the top US general said Tuesday.

US Air Force Gen. Dan Caine, chairman of the Joint Chiefs of Staff, gave an update on the US war in Iran on Tuesday morning and answered a question on the status of Iran’s air defense capabilities and the threat to American aircraft. “Most of their higher-end surface-to-air missile systems are not factors at this point,” he told reporters.

“We’re able to move around fighters that are moving deeper with relative impunity,” the general said.

The Department of Defense shared a map on Tuesday after the press briefing showing the timeline and targets for the first 10 days of Operation Epic Fury. US and Israeli strike points cover nearly all the designated “Iranian Air Defenses.”

In the early days of the fighting, Caine and other US officials pointed to Iranian air defenses and missiles as primary targets alongside Iran’s naval forces and nuclear sites.

The US and Israel have struck Iranian air defense systems with physical munitions, but separately, the Pentagon has indicated that cyber, space, and electronic warfare assets were used to blind Iranian defenses and disrupt their communications.

In October 2024, Israel carried out a significant strike targeting Iran’s missile production and air defenses, saying it could operate “with greater freedom in Iranian airspace.”

Israel conducted another preemptive strike in June 2025, destroying Iranian air defenses so its air forces could attack Iran’s nuclear facilities. Days later, the US carried out Operation Midnight Hammer aimed at destroying major Iranian nuclear sites.

After that operation, Caine said it didn’t look like Iran got a shot off at the strike package. “It appears that Iran’s surface-to-air missile systems did not see us,” he said.

He added that US fourth- and fifth-generation aircraft flew ahead of the B-2 Spirit bombers carrying massive 30,000-pound munitions, “sweeping in front of the package for enemy fighters and surface-to-air missiles.”

Air Force officials have said Iranian air defenses were unable to target US forces during the operation.F-35 Lightning II stealth fighters were tasked with suppressing air defenses.

Iran possesses a layered air defense network including a mix of its own systems and Russian ones, including the Bavar-373, an advanced, long-range system, and a variant of the Russian-made S-300.

At Monday’s press briefing, Caine said that with the degradation of Iran’s air defense network, aircraft can move with impunity. It’s an indication that the localized air superiority achieved early on in the fighting has expanded.

The danger isn’t totally removed though. “Always some risk out there,” Caine said. “I don’t want to say there isn’t.”

Since the war began late last month, the US and Israel have conducted over 5,000 strikes on targets. Nearly 2,000 of those were struck in the first 100 hours of the conflict, and the US has recorded major drops in the number of Iranian missiles and drones and the rate of fire.

• Mark Zuckerberg posted a photo to Threads with the Scale AI founder after rumors of tension.
• Meta has created a new AI engineering team, separate from Wang’s AI team at the company.
• Meta denies that Zuckerberg is “done with” Wang.

Folks, it’s just a totally normal day over at Meta HQ. Nothing to see here, no unsubstantiated rumors of a feud, no billion-dollar breakup. Just two totally happy, smiling colleagues.

See? Meta’s CEO, Mark Zuckerberg, just posted this photo of himself linking arms with Alexandr Wang, the founder of Scale AI and now head of AI Superintelligence at Meta. Just two happy dudes!

You, a normal person who happened to stumble across this photo while browsing Threads — where it’s been widely circulated — might be confused. What’s the deal with this photo?

Here’s the backstory:

Recently, The Wall Street Journal reported that Meta reorganized its AI structure, creating a new group of applied AI engineers. Meta had brought in Alexandr Wang as chief AI officer last summer, after investing more than $14 billion for a 49% stake in Wang’s data-labeling company, Scale AI. This brand new applied AI team, however, is separate from Wang’s team, and reports up to CTO Andrew Bosworth, the Journal reported.

Here’s where things get weird. Last week, the Times of India ran a headline saying “Mark Zuckerberg may be ‘done with’ Meta’s highest-paid employee as company’s reorganization shows.” That story went mildly viral on Reddit and was aggregated by IDN Financials. Meta’s spokesperson, Andy Stone, disputed the story on X, and as of Tuesday, the headline has been changed.

So is Zuckerberg’s photo designed to help quash rumors of tension between the two men? That Wang isn’t being pushed out, and they’re working together as normal?

While the whole blowup around the Times of India story was a big deal on X, where lots of people are interested in discussing news about the AI industry, Zuckerberg chose to post the photo to Threads, where chatter on AI office politics is much more muted.

Of course, Mark Zuckerberg has a lot of followers on Threads, but it seems most of them had no idea about what was going on. One of the top comments I saw was someone posting the “context?” guy meme, then more replies mistook Wang for Roy Lee of Cluely (oof), and someone replied asking Grok who was in the photo (oops).

And so here you have photographic evidence of two happy, normal guys — and now you know the backstory if you didn’t already. Now move along, nothing to see here!

• MrBeast is exploring offering advertising services, according to new trademark filings.
• The top YouTuber has filed applications that list plans for a range of ad and marketing services.
• It’s the latest sign of his ambitions to expand beyond his core media business.

YouTuber MrBeast broke the code for getting attention online. Now, he may be looking to share his strategies with others.

The content creator’s company filed a new set of trademark applications in the past few weeks that lay the groundwork for branching into advertising and marketing services.

Two of those filings, for “MrBeast Social” and “Beast Social,” include activities such as advertising, marketing, and promotion services, as well as facilitating connections between creators, brands, and agencies.

A third filing, under the name “Creator Industries,” lists plans for services for online content creators, like software and mobile apps for content creation, editing, monetization, and more.

The CEO of Beast Industries, Jeffrey Housenbold, said in December that the company planned to build a platform to connect brands and creators. Donaldson’s company already offers a “clipping” marketing service called Vyro, which matches brands with creators to promote video clips across social media. He also operates other content-related service businesses, such as the analytics company Viewstats and the dubbing company CreatorGlobal.

A fourth new trademark application, for “Watchtime Studios,” describes a video production arm focused on developing and producing video content for others.

MrBeast’s applications didn’t give a timeframe, but they were filed on an “intent to use” basis, which means the applicant has a bona fide intent to launch such services, said Josh Gerben, a trademark attorney whose firm’s clients include entrepreneurs, celebrities, and pro athletes, and who spotted the filings.

A spokesperson for MrBeast declined to comment.

Donaldson would be entering a crowded field in advertising and content creator services. The competition includes ad holding companies, tech platforms, and independent agencies. On the other hand, brands and creators could be drawn to the idea of the world’s biggest online creator helping them master social media themselves. For example, Donaldson recently worked with Salesforce to develop a concept for its Super Bowl commercial, which he also appeared in.

Other entertainers have entered the advertising and marketing space lately, most notably Ryan Reynolds through his agency, Maximum Effort. Other recent examples are podcaster Alex Cooper, who in 2025 launched the Unwell Creative Agency, a division of her media company Unwell, and Kristen Bell, who launched Dunshire Productions in 2022.

Two former MrBeast staffers, Anthony Potero and Talia Schulhof, also launched a stunt-marketing startup called Pufferfish last year.

MrBeast has been branching out into new businesses beyond media as he tries to improve the company’s financial footing ahead of a possible IPO.

In the past year, he’s acquired Step, a financial app for teens; explored a mobile phone service in the vein of Ryan Reynolds’ Mint Mobile; and dipped a toe into experiences with Beast Land, a pop-up theme park in Saudi Arabia. He also has a growing candy business, Feastables.

In January, Beast Industries, which was previously valued at around $5 billion, announced it raised $200 million in new funding from the ethereum holding company Bitmine Immersion Technologies.

• The Pentagon published an awkwardly worded press release announcing a soldier’s death.
• The press release is titled “DOW Identifies An Army Believed to Be Casualty.”
• Retired military spokespeople say casualty announcements are among the military’s most sensitive communications.

The Pentagon last week announced the death of a US Army soldier killed while supporting Operation Epic Fury before a medical examiner had positively identified them. Former military spokespeople said that it was an unusual and awkwardly phrased departure from standard procedures.

In a press release titled, “DOW Identifies An Army Believed to Be Casualty,” the Department of Defense announced “the believed to be death” of Chief Warrant Officer 3 Robert M. Marzan, an Army reservist who died during an Iranian strike that also killed five other troops in Kuwait.

According to the release, Marzan, a soldier with the 103rd Sustainment Command, “was at the scene of the incident on March 1, 2026, and is believed to be the individual who perished at the scene. Positive identification of Chief Warrant Officer 3 Marzan will be completed by the medical examiner.”

A defense official told Business Insider a medical examiner has since confirmed Marzan’s identity.

Prematurely announcing a death risks misidentification, which can erode public trust if corrections are later required, two former military spokespeople told Business Insider. They also said attention to detail and clarity in these communications shows respect.

“When a service member is killed in combat, they deserve better than this,” Joe Plenzler, a retired Marine lieutenant colonel who worked in public affairs during the Global War on Terror, wrote in a post to LinkedIn.

He told Business Insider separately that “it’s a simple matter of respect to make sure that everything is accurate.”

US Central Command, which oversees American military operations in the Middle East, has reported seven US service members killed in action in the Iran war that began in late February. The fatal strike in Kuwait came as Iranian forces launched missiles into countries across the region. Marzan is the only service member in this conflict so far who has been described as “believed to be” dead.

Asked about the statement, the Office of the Secretary of Defense referred Business Insider to the Army, saying, “DOW announces, all follow-on questions go to the Army.” The Army did not provide comment to Business Insider.

Why the Pentagon statement was unusual

Typically, the military refers to a service member whose death has not yet been confirmed as “DUSTWUN,” short for “duty status — whereabouts unknown,” a retired Army spokesman who served during the wars in Iraq and Afghanistan told Business Insider. The term is used when a service member’s absence is involuntary and their status cannot yet be confirmed.

Announcing a death before positive identification by a medical examiner marks a break from norms that governed casualty reporting over two previous decades of war, the former spokesman said. The DUSTWUN designation is intended for situations where ongoing rescue efforts prevent an immediate determination, though recovery of remains is not always required to declare a service member deceased, according to military policies outlining casualty procedures.

“We had thousands of casualties throughout the conflicts in Iraq and Afghanistan and Syria,” said the retired Army official. “I don’t recall ever announcing someone as ‘believed to be a casualty.'”

Few communications are as important or sensitive as announcing a casualty, he said, describing a somber process honed after more than 7,000 US service member deaths during the Global War on Terror, according to Brown University’s Costs of War project.

Plenzler, the former Marine public affairs officer, told Business Insider that all communications related to sensitive topics, including casualties, were generally examined by at least three people before publication because of the heavy impact on public trust.

In his LinkedIn post, he recalled seeing “people removed from leadership positions for getting names incorrect during memorial services.”

While the former Army spokesman expressed disappointment in what they characterized as an awkwardly written DoD announcement, he also noted that many of the personnel who oversaw casualty communications during the height of the previous wars in the Middle East have since left the service, leaving newer troops to manage hard notifications and public messaging.

“We have been sort of out of this business now for several years,” he said.

Marzan, 54, lived in Sacramento and was assigned to an Iowa-based logistics unit. Business Insider could not reach a Marzan family member for comment.

Communicating casualty updates comes with a learning curve, the former Army spokesman said. The details of this release are unclear, but he said he hopes “they’ve learned a lesson from this.”

The announcement comes amid broader shifts in how the military communicates during fast-moving combat operations, including increased reliance on social media updates from combatant commands and the Pentagon. Communications once known for staid military-speak now often feature videos of US missile strikes or jets taking off, strong wartime rhetoric, or posts debunking Iranian “bogus claims.”