Every co-founder Elon Musk recruited to build xAI has now reportedly left the company. Manuel Kroiss, who led the pretraining team, told people this month that he was departing. Ross Nordeen, described by Business Insider as Musk’s “right-hand operator,” left on Friday. They were the last two of eleven co-founders, all of whom have exited a company that was valued at $250 billion when SpaceX acquired it in February and that Musk himself described two weeks ago as having been “not built right the first time around.”

The departures are not ordinary startup attrition. The researchers Musk assembled in 2023 were among the most accomplished in artificial intelligence. Jimmy Ba co-authored the 2014 Adam optimisation paper, the most-cited paper in AI with more than 95,000 citations. Igor Babuschkin, the chief engineer, came from Google DeepMind. Christian Szegedy came from Google. Tony Wu led the reasoning team. Greg Yang, Toby Pohlen, Zihang Dai, Guodong Zhang, and Kyle Kosic brought experience from DeepMind, Google, Microsoft, and OpenAI. That entire cohort is now gone, and the company they helped build is being, in Musk’s words, “rebuilt from the foundations up.”

A timeline of unravelling

The exodus accelerated sharply in early 2026. Christian Szegedy left in February 2025, an early signal. But the cascade began in earnest when Tony Wu, one of the most operationally central co-founders, announced his departure on February 10, 2026. Jimmy Ba resigned within 24 hours, reportedly amid tensions over demands to improve model performance. By mid-March, only Kroiss and Nordeen remained. Their departures this week complete the sweep.

The timing is difficult to separate from the corporate restructuring happening around xAI. On February 2, SpaceX acquired xAI in an all-stock transaction that valued SpaceX at $1 trillion and xAI at $250 billion, creating a combined entity worth $1.25 trillion, the largest corporate merger by valuation in history. The deal brought xAI, X (formerly Twitter), and SpaceX under a single corporate umbrella, with SpaceX now preparing for a potential IPO in mid-2026 that could target a $1.75 trillion valuation.

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

Weeks earlier, in January, Tesla invested $2 billion in xAI’s Series E round at an approximate $230 billion valuation. Tesla shareholders are suing Musk for breach of fiduciary duty over the investment, arguing that the company’s chief executive effectively directed shareholder capital into his own private venture. The lawsuit gained additional force on March 13, when Musk publicly acknowledged that xAI’s products, particularly its coding tools, were not competitive with Anthropic’s Claude Code or OpenAI’s Codex. Tesla had invested $2 billion in a company whose founder admitted it needed to be rebuilt from scratch.

What “not built right” means at $250 billion

Musk’s admission on March 13 was unusually candid for a chief executive whose company had just been acquired for a quarter of a trillion dollars. He said xAI’s AI coding tools simply did not work, and that the underlying system needed to be rebuilt. The statement appeared to validate the co-founders’ decision to leave: if the company’s own leadership acknowledges that the product failed, the researchers who built it have limited incentive to stay for the rebuild, particularly when they can command extraordinary compensation at competitors.

The AI talent market in 2026 is the most competitive it has ever been. Meta has reportedly offered packages worth up to $300 million over four years to retain top AI researchers. OpenAI, Google DeepMind, and Anthropic are all expanding their research teams aggressively. The eleven researchers who left xAI represent a concentration of talent that any of those companies would pay handsomely to acquire. Where they end up will say as much about the industry’s future direction as their departure says about xAI’s past.

xAI is not without assets. The Colossus supercomputer, built with more than 200,000 NVIDIA H100 GPUs, remains one of the largest AI training clusters in the world. Grok, the company’s chatbot, has a distribution channel through X’s user base. And the SpaceX merger provides access to capital, infrastructure, and engineering talent at a scale that few AI companies can match. The question is whether infrastructure and distribution are sufficient when the research leadership that was supposed to make the product competitive has entirely departed.

The pattern

The xAI co-founder exodus follows a pattern that has repeated across Musk’s companies. Twitter lost the majority of its senior leadership and roughly 80 per cent of its workforce within months of his 2022 acquisition. Tesla’s senior ranks have thinned steadily as Musk’s attention has divided across six companies. The common thread is a management style that produces extraordinary results in hardware engineering, where Musk’s tolerance for risk and pace of iteration have built SpaceX and Tesla into industry-defining companies, but appears less effective in research-driven fields where the most valuable people have abundant alternatives and low tolerance for instability.

Artificial intelligence research is, in 2026, the most competitive labour market in technology. The researchers who co-founded xAI did not need to be there. They chose to be, attracted by the resources Musk could deploy and the ambition of the project. That every one of them has now chosen to leave, during a period when the company received a $250 billion valuation and access to the resources of SpaceX, suggests that the problems at xAI are not principally financial or infrastructural. They are organisational. And no amount of capital can rebuild a research culture once the people who created it have gone.

The air felt different at this year’s SXSW, the annual March festival where tech meets pop culture in Austin. I was reminded of the 2019 SXSW when people packed downtown, and snake lines formed out of local ventures. 

Attendees said it was like that again this year, though my friend, who lives in the area and has attended many times, admitted that some stuff has changed. For instance the festival is now two days shorter than it used to be. It was also “decentralized,” mainly due to the demolition of the Austin Convention Center, which scattered events and panels throughout downtown venues. That made the whole conference feel less overwhelming but also less connected.

The event is also still recovering from the pandemic, during which it laid off staff and went two years without much income. It’s switched hands since then and, as of this year, has adopted a new strategy.

Greg Rosenbaum, the SVP of programming at SXSW, said this year, the conference’s 40th anniversary, was its most “ambitious reinvention” yet. He cited changes like the new Clubhouses, for recharging, networking, and special programming, that attracted 5,000 people daily. He noted how attendees were experiencing “more of Austin and the downtown community.”

For at least the tech founders I spoke with, the conference remains immensely valuable, and everyone had the same advice: conferences like these, you get what you give. 

After all, there were people to meet and panels to speak on. The Grammy-nominated Lola Young performed, Vox threw a hot party, the new Boots Riley film premiered, while Serena Williams and Steven Spielberg had keynotes. (I also moderated a panel about AI and taboo topics like relationships and money, which was pretty good if you ask me.) 

Ashley Tryner-Dolce, an investor and founder, said the conference was still an “incredible gathering of ideas.” Like many festivals, though, she found the most “meaningful moments” happened at the side events — like INC’s Founder House party, where she connected with other founders and CEOs. 

“It’s less about the main stage and more about who you’re sitting across from,” she said. 

James Norman, a managing partner at Black Ops VC, didn’t even have a proper badge to the festival. He threw an event to connect founders with opportunities and attended some film screenings and dinners.

“If you’re just showing up without the right connections or proximity to the rooms and conversations that matter, you’re going to struggle to unlock the real value of the event,” he said, which is exactly what Jonathan Sperber, a founder who participated in the SXSW pitch competition, also expressed. 

“The value tends to depend on how well you prepare for it,” Sperber said, adding that his team made sure to have meetings lined up and a clear strategy going in. He called it an “effective setting for connecting with large enterprises and other key stakeholders.” 

The talk of SXSW being dead has circled the industry for years, but that never seems to be the case. For every batch of tiring founders, emerges a crop of fresh eyes and ambition, ready to take advantage of what lies in the festival’s wake. 

For example, this was Simon Davis’ first SXSW. He said that his overall impression was that it was “a media conference with a tech angle, not the other way around.” He praised the diversity of the event compared to other tech events (which we will spare to mentioning).

“At SXSW, you get a much wider range of people, backgrounds, and experience levels,” he continued. “The live music programming reinforces that. It’s a different energy entirely. Not somewhere you’d necessarily go to do deals as a tech company, but a great place to share and learn.” 

This year, SXSW introduced a new badging system, meaning each person had a different experience, depending on what track badge they bought — film, music, or tech. I, for example, felt surrounded by conversations about AI and technology, and overheard other tech people talking about how the festival once had a stronger music focus (though it did seem, for sure, that there were more tech-focused panels this year than music showcases or film opportunities).

The conference also eliminated the secondary access that let people with, say, music badges get into film events. Instead, people had to buy the all-in-one premium badge for around $2,000. It also introduced a reservation system (to help with lines), where badge holders had to book time for whatever they wanted to do. That was true even for those with a platinum badge, like Sperber. 

As a result, he said the festival didn’t feel like a place where anyone could just show up, and noted that some events booked up so quickly they were difficult to get into. The decentralized bit also made it harder to get around than he would have liked. 

“I liked the openness and the ability to meet folks from all life experiences, got to really understand the city, and some of the interactive exhibits were very interesting,” he said. 

Rosenbaum said the team made the decision to get rid of secondary access after hearing feedback that attendees want more of a “streamlined access across the badges, as well as more benefits for Platinum badges.” They also lowered the price of the platinum badge to make the all-in-one option more affordable. Reservations, meanwhile, will return next year, he said, citing positive feedback (aside from a few technical errors and capacity confusion). “We will certainly adjust and refine them as needed,” he said.

Norman described it as more of an “unconference” now, at least from his perspective. He said the event was more flexible, allowing people to move around, meet people, and then go to other places. 

Rodney Williams, the co-founder of the fintech SoLo Funds, has also noticed a change, but again, it’s not necessarily a bad one. He’s been going to SXSW for more than a decade and has hosted events and spoken on panels. Usually, he goes for the entire festival, but this year, he decided to go only for a few days, throwing his own events and avoiding lines.  

He said that for tech founders, SXSW has “moved from an intimate, scrappy discovery zone to a high-cost, high-competition space,” focused on “investor interaction and experiential marketing” — meaning companies with big budgets can put on the big activations and get more eyeballs. 

“If you are attending for the first time or don’t have access to the right events or connections, the event can definitely prove to be tricky,” Williams said. 

Adweek reported fewer spectacles overall and said that there was an absence of big tech companies advertising. Williams elucidated that even with the lack of big tech companies, advertising is still a big-bucks game.

“Companies with massive marketing budgets are usually the only ones participating, launching products, or throwing pricey events,” he said. “It wasn’t always like this, and that shift has taken away opportunities from the emerging tech companies that used to participate.”

Williams added, “Now, standing out requires more than just a great product, demanding significant marketing investment that only companies with huge budgets can do.” 

That didn’t stop him from throwing a party this year. Norman either. In fact, the organizers expected around 300,000 people to show up this year (final numbers won’t be available until April), revealing that the conference has yet to lose its steam or its magic. 

“I always enjoy it and make the most out of it,” Williams said.

A vulnerability in the Smart Slider 3 WordPress plugin, active on more than 800,000 websites, can be exploited to allow subscriber-level users access to arbitrary files on the server.

An authenticated attacker could use it to access sensitive files, such as wp-config.php, which includes database credentials, keys, and salt data, creating the risk for user data theft and complete website takeover.

Smart Slider 3 is one of the most popular WordPress plugins for creating and managing image sliders and content carousels. It offers an easy-to-use drag-and-drop editor and a rich set of templates to choose from.

The security issue, tracked as CVE-2026-3098, was discovered and reported by researcher Dmitrii Ignatyev and impacts all versions of the Smart Slider 3 plugin through 3.5.1.33.

It received a medium severity score due to requiring authentication. However, this only limits the impact to websites with membership or subscription options, a feature that is common on many platforms these days.

The vulnerability stems from missing capability checks in the plugin’s AJAX export actions. This allows any authenticated user, including subscribers, to invoke them.

According to researchers at WordPress security company Defiant, the developer of the Wordfence security plugin, the ‘actionExportAll’ function lacks file type and source validation, thus allowing arbitrary server files to be read and added to the export archive.

The presence of a nonce does not prevent abuse because it can be obtained by authenticated users.

“Unfortunately, this function does not include any file type or file source checks in the vulnerable version. This means that not only image or video files can be exported, but .php files can as well,” says István Márton, a vulnerability research contractor at Defiant.

“This ultimately makes it possible for authenticated attackers with minimal access, like subscribers, to read any arbitrary file on the server, including the site’s wp-config.php file, which contains the database credentials as well as keys and salts for cryptographic security.”

500K websites still vulnerable

On February 23, Ignatyev reported his findings to Wordfence, whose researchers validated the provided proof-of-concept exploit and informed Nextendweb, the developer of Smart Slider 3.

Nextendweb acknowledged the report on March 2 and on March 24 delivered a patch with the release of Smart Slider version 3.5.1.34.

According to WordPress.org stats, the plugin was downloaded 303,428 times over the past week. This means that at least 500,000 WordPress sites are running a vulnerable version of the Smart Slider 3 plugin and are exposed to attacks.

CVE-2026-3098 is not flagged as actively exploited as of writing, but the status may change soon, so prompt action is required by website owners/administrations.

Automated pentesting proves the path exists. BAS proves whether your controls stop it. Most teams run one without the other.

This whitepaper maps six validation surfaces, shows where coverage ends, and provides practitioners with three diagnostic questions for any tool evaluation.

Get Your Copy Now