Elon Musk and Sam Altman are going to court over OpenAI’s future

After a yearslong legal feud, Elon Musk and OpenAI CEO Sam Altman are heading to trial this week in Northern California in a case that could have sweeping consequences. Ahead of OpenAI’s highly anticipated IPO, the court could rule on whether the company is allowed to exist as a for-profit enterprise and might even oust its current executive leadership, including Altman.

Musk is suing OpenAI, alleging that Altman and OpenAI president Greg Brockman deceived him into bankrolling the company in its early days by promising to maintain it as a nonprofit dedicated to developing AI that benefits humanity, only to later restructure the company to operate a for-profit subsidiary. Musk cofounded OpenAI with Altman and others in 2015, but he left in 2018 after a bitter power struggle. 

Musk is seeking as much as $134 billion in damages from OpenAI and Microsoft, one of OpenAI’s biggest financial backers. He is also asking the court to remove Altman and Brockman from their roles and to restore OpenAI as a nonprofit. Musk has asked the court to award any damages to OpenAI’s nonprofit rather than to him personally. 

Nine jurors will deliver an advisory verdict, a non-binding recommendation, to guide the judge in deciding Musk’s claims against Altman. Musk, Altman, and Brockman will take the stand. Former OpenAI chief scientist Ilya Sutskever, former OpenAI CTO Mira Murati, and Microsoft CEO Satya Nadella are also expected to testify. Cringey texts, raw diary entries, and endless scheming behind the founding and growth of OpenAI are expected to come to light.

In an industry enveloped in secrecy, the trial will be a rare opportunity for the public to look behind the curtain and find out what’s going on in the companies creating the most transformative technology ever built. 

What are they fighting about?

When OpenAI was originally founded as a nonprofit, backed by a $38 million donation from Musk, the company vowed to create open-source technology for the public’s benefit, unconstrained by a need to generate financial returns. But over the years, the company began to claim that intensifying competition could make it dangerous to share how it develops its AI models and that a nonprofit structure could not raise enough money to keep building AI. (MIT Technology Review was first to report on OpenAI’s internal conflicts around its mission.)

The court has already found that in 2017 Altman and Brockman wanted to establish a for-profit arm, while Musk proposed merging OpenAI with his electric-car company, Tesla. When Musk threatened to stop funding, Altman and Brockman told him that they were committed to keeping the company a nonprofit. Musk alleges that they pursued plans to pivot to a for-profit without informing him. According to OpenAI, Musk agreed that the company needed a for-profit entity and even wanted to be its CEO. 

But even if Musk proves he was duped by Altman and Brockman, he may not have standing in the first place to sue them for restructuring the company to operate a for-profit subsidiary. Some legal scholars are puzzled over why the judge allowed him to bring this claim. “The idea that Elon Musk can sue because he was a donor or used to be on the board is pretty puzzling,” says Jill Horwitz, a law professor who studies nonprofit law at Northwestern University. “Typically, it’s up to the attorneys general to bring such a claim to enforce the charitable purposes. And that’s already happened.” 

In October 2025, state attorneys general of California, where OpenAI is headquartered, and Delaware, where OpenAI is incorporated, struck a deal with OpenAI to approve its new corporate structure on a series of conditions. For example, a safety and security committee at the nonprofit would review safety-related decisions made by the for-profit subsidiary. Critics of the restructuring, including Musk, AI safety advocates, and civil society groups, have tried to stop it. 

California’s attorney general has declined to join Musk’s lawsuit, saying that the office did not see how his action serves the public interest.

Still, whether the deal holds OpenAI to its nonprofit mission is an open question. “Elon Musk should have to show … what the deficiencies are in what’s been agreed to by OpenAI with the attorneys general,” says Rose Chan Loui, the director of the UCLA School of Law’s philanthropy and nonprofit program. Even with the terms in place, holding OpenAI to them depends on “how much they can enforce it and how much transparency they get into OpenAI’s work.”

More importantly, legal experts say the case is being considered under the wrong body of law. Musk argues that Altman and Brockman breached OpenAI’s charitable trust by creating a closed-source, for-profit subsidiary. As a result, the court has been analyzing the claim under the law of trusts. “But OpenAI is not a trust. OpenAI is a corporation. And so really they should be looking at … the law of charitable nonprofit organizations,” says Chan Loui.

What’s on the line?

Despite all the legal muddiness, the outcome of the trial could upend the AI race. Any one of the remedies that Musk seeks could cripple OpenAI as it races to go public by the end of the year. OpenAI, which is valued at over $850 billion, has described the litigation with Musk as a potential risk to its business. Musk’s rival company xAI, which makes the chatbot Grok, is expected to go public as a part of his rocket company SpaceX as early as June. If Musk prevails, xAI, which in combination with SpaceX is valued at $1.25 trillion, could get a big advantage in the AI race. 

And the trial has helped expose the bitter schism between Musk and the company he once helped to found. An OpenAI spokesperson referred MIT Technology Review to a post on X: “This lawsuit has always been a baseless and jealous bid to derail a competitor.” Although Musk’s lawyers did not immediately respond to a request for comment, he has posted on X that “Scam Altman lies as easily as he breathes.”  

MIT Technology Review will have ongoing coverage of Musk v. Altman until its conclusion. Follow @techreview or @michelletomkim on X for up-to-the-minute reporting. 

Three reasons why DeepSeek’s new model matters

On Friday, Chinese AI firm DeepSeek released a preview of V4, its long-awaited new flagship model. Notably, the model can process much longer prompts than its last generation, thanks to a new design that helps it handle large amounts of text more efficiently. Like DeepSeek’s previous models, V4 is open source, meaning it is available for anyone to download, use, and modify.

V4 marks DeepSeek’s most significant release since R1, the reasoning model it launched in January 2025. R1, which was trained on limited computing resources, stunned the global AI industry with its strong performance and efficiency, turning DeepSeek from a little-known research team into China’s best-known AI company almost overnight. It also helped set off a wave of open-weight model releases from other Chinese AI firms. 

DeepSeek has kept a relatively low profile since then—but earlier this month, it effectively teased V4’s release when it added “expert” and “flash” modes to the online version of its model, prompting speculation that the updates were tied to a bigger upcoming release.

While the company has become a powerful symbol of China’s AI ambitions, its big return to cutting-edge frontier models comes after months of scrutiny—including major personnel departures, delays to previous model launches, and growing scrutiny from both the US and Chinese governments. 

So, will V4 shake the AI field the way R1 did? Almost certainly not, but here are three big reasons why this release matters.

1. It breaks new ground for an open-source model.

As with R1 before it, DeepSeek claims that V4’s performance rivals the best models available at a fraction of the price. This is great news for developers and for companies using the tech, because it means they can access frontier AI capabilities on their own terms, and without worrying about skyrocketing costs.

The new model comes in two versions, both of which are available on DeepSeek’s website and in its app, with API access also open to developers. V4-Pro is a larger model built for coding and complex agent tasks, and V4-Flash is a smaller version designed to be faster and cheaper to run. Both versions offer reasoning modes, in which the model can carefully parse a user’s prompt and show each step as it works through the problem.

For V4-Pro, DeepSeek charges $1.74 per million input tokens and $3.48 per million output tokens, a fraction of the cost of comparable models from OpenAI and Anthropic. V4-Flash is even cheaper, at about $0.14 per million input tokens and about $0.28 per million output tokens, making it one of the cheapest top-tier models available. This would make it a very appealing model to build applications on.

In terms of performance, V4 is, perhaps unsurprisingly, a huge jump from R1—and it seems to be a strong alternative to just about all the latest big AI models. On the major benchmarks, according to results shared by the company, DeepSeek V4-Pro competes with leading closed-source models, matching the performance of Anthropic’s Claude-Opus-4.6, OpenAI’s GPT-5.4, and Google’s Gemini-3.1. And compared to other open-source models, such as Alibaba’s Qwen-3.5 or Z.ai’s GLM-5.1, DeepSeek V4 exceeds them all on coding, math, and STEM problems, making it one of the strongest open-source models ever released. 

DeepSeek also says that V4-Pro now ranks among the strongest open-source models on benchmarks for agentic coding tasks and performs well on other tests that measure ability to carry out multistep problems. Its writing ability and world knowledge also lead the field, according to benchmarking results shared by the company. 

In a technical report released alongside the model, DeepSeek shared results from an internal survey of 85 experienced developers: More than 90% included V4-Pro among their top model choices for coding tasks.

DeepSeek says it has specifically optimized V4 for popular agent frameworks such as Claude Code, OpenClaw, and CodeBuddy.

2. It delivers on a new approach to memory efficiency.

One of the key innovations of V4 is its long context window—the amount of text the model can process at once. Both versions can handle 1 million tokens, which is large enough to fit all three volumes of The Lord of the Rings and The Hobbit combined. The company says this context window size is now the default across all DeepSeek services and it matches what is offered by cutting-edge versions of models like Gemini and Claude. 

But it’s important to know not just that DeepSeek has made this leap, but how it did so. V4 makes significant architectural changes to the company’s former models—especially in the attention mechanism, which is the feature of AI models that helps them understand each part of a prompt in relation to the rest. As the prompt text gets longer, these comparisons become much more costly, making attention one of the main bottlenecks for long-context models.

DeepSeek’s innovation was to make the model more selective about what it pays attention to. Instead of treating all earlier text as equally important, V4 compresses older information and focuses on the parts most likely to matter in the present moment, while still keeping nearby text in full so it does not miss important details. 

DeepSeek says this sharply reduces the cost of using long context. In a 1-million-token context, V4-Pro uses only 27% of the computing power required by its previous model, V3.2, while cutting memory use to 10%. The reduction in V4-Flash is even larger, using just 10% of the computing power and 7% of the memory. In practice, this could make it cheaper to build tools that need to work across huge amounts of material, such as an AI coding assistant that can read an entire codebase or a research agent that can analyze a long archive of documents without constantly forgetting what came before.

DeepSeek’s interest in long context windows didn’t start with V4. Over the past year and a half, the company has quietly published a series of papers on how AI models “remember” information, experimenting with compression and mathematical techniques to extend what AI models could realistically handle.

3. It marks the first steps on the hard road away from Nvidia.

V4 is DeepSeek’s first model optimized for domestic Chinese chips, such as Huawei’s Ascend—a move that has turned the launch into something of a test of whether China’s homegrown AI industry can begin to loosen its dependence on US chip giant Nvidia. 

This was largely expected, since The Information reported earlier this month that DeepSeek did not give American chipmakers like Nvidia and AMD early access to V4, though prerelease access is common to allow chipmakers to optimize support of the new model ahead of a launch. Instead, the company reportedly gave early access only to Chinese chipmakers. 

On Friday, Huawei said its Ascend supernode products, based on the Ascend 950 series, would support DeepSeek V4. This means that companies and individuals who want to run their own modified version of Deepseek V4 will be able to use Huawei chips easily.

Reuters previously reported that Chinese government officials recommended that DeepSeek integrate Huawei chips in its training process. And this pressure fits a broader pattern in China’s industrial policy: Strategic sectors are often pushed, and sometimes effectively required, to align with national self-reliance goals. But there’s a particular urgency when it comes to AI. Since 2022, US export controls have cut Chinese firms off from Nvidia’s most powerful chips, and they later also restricted access to downgraded China-market versions. Beijing’s response has been to accelerate the push for a domestic AI stack, from chips to software frameworks to data centers.

Chinese authorities have reportedly been pushing data centers and public computing projects to use more domestic chips, including through reported bans on foreign-made chips, sourcing quotas, and requirements to pair Nvidia chips with Chinese alternatives from companies such as Huawei and Cambricon. 

Still, replacing Nvidia is not as simple as swapping one chip for another. Nvidia’s advantage lies not only in its chips, but in the software ecosystem developers have spent years building around them. Moving to Huawei’s Ascend chips means adapting model code, rebuilding tools, and proving that systems built around those chips are stable enough for serious use.

To be clear, DeepSeek does not appear to have fully moved beyond Nvidia. The company’s technical report reveals that it is using Chinese chips to run the model for inference, or when someone asks the model to complete a task. But Liu Zhiyuan, a computer science professor at Tsinghua University, told MIT Technology Review that DeepSeek appears to have adapted only part of V4’s training process for Chinese chips. The report does not say whether some key long-context features were adapted to domestic chips, so Liu says V4 may still have been trained mainly on Nvidia chips. Multiple sources who spoke on the condition of anonymity, due to political sensitivity around these issues, told MIT Technology Review that Chinese chips still don’t perform as well as Nvidia chips but are better suited for inference than training.

DeepSeek is also tying the future costs of V4 to this hardware shift. The company says V4-Pro prices could fall significantly after Huawei’s Ascend 950 supernodes begin shipping at scale in the second half of this year. 

If that works, V4 could be an early sign that China is successfully building a parallel AI infrastructure.

Chinese tech workers are starting to train their AI doubles–and pushing back

Tech workers in China are being instructed by their bosses to train AI agents to replace them—and it’s prompting a wave of soul-searching among otherwise enthusiastic early adopters. 

Earlier this month a GitHub project called Colleague Skill, which claimed workers could use it to “distill” their colleagues’ skills and personality traits and replicate them with an AI agent, went viral on Chinese social media. Though the project was created as a spoof, it struck a nerve among tech workers, a number of whom told MIT Technology Review that their bosses are encouraging them to document their workflows in order to automate specific tasks and processes using AI agent tools like OpenClaw or Claude Code. 

To set up Colleague Skill, a user names the coworker whose tasks they want to replicate and adds basic profile details. The tool then automatically imports chat history and files from Lark and DingTalk, both popular workplace apps in China, and generates reusable manuals describing that coworker’s duties—and even their unique quirks—for an AI agent to replicate. 

Colleague Skill was created by Tianyi Zhou, who works as an engineer at the Shanghai Artificial Intelligence Laboratory. Earlier this week he told Chinese outlet Southern Metropolis Daily that the project was started as a stunt, prompted by AI-related layoffs and by the growing tendency of companies to ask employees to automate themselves. He didn’t respond to requests for further comment.

Internet users have found humor in the idea behind the tool, joking about automating their coworkers before themselves. However, Colleague Skill’s virality has sparked a lot of debate about workers’ dignity and individuality in the age of AI.

After seeing Colleague Skill on social media, Amber Li, 27, a tech worker in Shanghai, used it to recreate a former coworker as a personal experiment. Within minutes, the tool created a file detailing how that person did their job. “It is surprisingly good,” Li says. “It even captures the person’s little quirks, like how they react and their punctuation habits.” With this skill, Li can use an AI agent as a new “coworker” that helps debug her code and replies instantly. It felt uncanny and uncomfortable, Li says. 

Even so,  replacing coworkers with agents could become a norm. Since OpenClaw became a national craze, bosses in China have been pushing tech workers to experiment with agents. 

Although AI agents can take control of your computer, read and summarize news, reply to emails, and book restaurant reservations for you, tech workers on the ground say their utility has so far proven to be limited in business contexts. Asking employees to make manuals describing the minutiae of their day-to-day jobs the way Colleague Skill does is one way to help bridge that gap. 

Hancheng Cao, an assistant professor at Emory University who studies AI and work, believes that companies have good reasons to push employees to create work blueprints like these, beyond simply following a trend. “Firms gain not only internal experience with the tools, but also richer data on employee know-how, workflows, and decision patterns. That helps companies see which parts of work can be standardized or codified into systems, and which still depend on human judgment,” he says.

To employees, though, making agents or even blueprints for them can feel strange and alienating. One software engineer, who spoke with MIT Technology Review anonymously because of concerns about their job security, trained an AI (not Colleague Skill) on their workflow and found that the process felt reductive—as if their work had been flattened into modules in a way that made them easier to replace. On social media, workers have turned to bleak humor to express similar feelings. In one comment on Rednote, a user wrote that “a cold farewell can be turned into warm tokens,” quipping that if they use Colleague Skill to distill their coworkers into tasks first, they themselves might survive a little longer.

The push for creating agents has also spurred clever countermeasures. Irritated by the idea of reducing a person to a skill, Koki Xu, 26 an AI product manager in Beijing, published an “anti-distillation” skill on GitHub on April 4. The tool, which took Xu about an hour to build, is designed to sabotage the process of creating workflows for agents. Users can choose between light, medium, and heavy sabotage modes depending on how closely their boss is observing the process, and the agent rewrites the material into generic, non-actionable language that would produce a less useful AI stand-in. A video Xu posted about the project went viral, drawing more than 5 million likes across platforms.

Xu told MIT Technology Review that she has been following the Colleague Skill trend from the start and that it has made her think about alienation, disempowerment, and broader implications for labor. “I originally wanted to write an op-ed, but decided it would be more useful to make something that pushes back against it,” she says.

Xu, who has undergraduate and master’s degrees in law, said the trend also raises legal questions. While a company may be able to argue that work chat histories and materials created on a work laptop are corporate property, a skill like this can also capture elements of personality, tone, and judgment, making ownership much less clear. She said she hopes Colleague Skill prompts more discussion about how to protect workers’ dignity and identity in the age of AI. “I believe it’s important to keep up with these trends so we (employees) can participate in shaping how they are used,” she says. Xu herself is an avid AI adopter, with seven OpenClaw agents set up across her personal and work devices.

Li, the tech worker in Shanghai, says her company has not yet found a way to replace actual workers with AI tools, largely because they remain unreliable and require constant supervision. “I don’t feel like my job is immediately at risk,” she says. “But I do feel that my value is being cheapened, and I don’t know what to do about it.”

How robots learn: A brief, contemporary history

Roboticists used to dream big but build small. They’d hope to match or exceed the extraordinary complexity of the human body, and then they’d spend their career refining robotic arms for auto plants. Aim for C-3P0; end up with the Roomba. 

The real ambition for many of these researchers was the robot of science fiction—one that could move through the world, adapt to different environments, and interact safely and helpfully with people. For the socially minded, such a machine could help those with mobility issues, ease loneliness, or do work too dangerous for humans. For the more financially inclined, it would mean a bottomless source of wage-free labor. Either way, a long history of failure left most of Silicon Valley hesitant to bet on helpful robots.

Imagine you’d like a pair of robot arms installed in your home purely to do one thing: fold clothes. How would it learn to do that? You could start by writing rules. Check the fabric to figure out how much deformation it can tolerate before tearing. Identify a shirt’s collar. Move the gripper to the left sleeve, lift it, and fold it inward by exactly this distance. Repeat for the right sleeve. If the shirt is rotated, turn the plan accordingly. If the sleeve is twisted, correct it. Very quickly the number of rules explodes, but a complete accounting of them could produce reliable results. This was the original craft of robotics: anticipating every possibility and encoding it in advance.

Around 2015, the cutting edge started to do things differently: Build a digital simulation of the robotic arms and the clothes, and give the program a reward signal every time it folds successfully and a ding every time it fails. This way, it gets better by trying all sorts of techniques through trial and error, with millions of iterations—the same way AI got good at playing games.

The arrival of ChatGPT in 2022 catalyzed the current boom. Trained on vast amounts of text, large language models work not through trial and error but by learning to predict what word should come next in a sentence. Similar models adapted to robotics were soon able to absorb pictures, sensor readings, and the position of a robot’s joints and predict the next action the machine should take, issuing dozens of motor commands every second.

This conceptual shift—to reliance on AI models that ingest large amounts of data—seems to work whether that helpful robot is supposed to talk to people, move through an environment, or even do complicated tasks. And it was paired with other ideas about how to accomplish this new way of learning, like deploying robots even if they aren’t yet perfect so they can learn from the environment they’re meant to work in. Today, Silicon Valley roboticists are dreaming big again. Here’s how that happened. 

Jibo

Jibo

A movable social robot carried out conversations long before the age of LLMs.

An MIT robotics researcher named Cynthia Breazeal introduced an armless, legless, faceless robot called Jibo to the world in 2014. It looked, in fact, like a lamp. Breazeal’s aim was to create a social robot for families, and the idea pulled in $3.7 million in a crowdsourced funding campaign. Early preorders cost $749.

The early Jibo could introduce itself and dance to entertain kids, but that was about it. The vision was always for it to become a sort of embodied assistant that could handle everything from scheduling and emails to telling stories. It earned a number of devoted users, but ultimately the company shut down in 2019.

COURTESY OF MIT MEDIA LAB

In retrospect, one thing that Jibo really needed was better language capabilities. It was competing against Apple’s Siri and Amazon’s Alexa, and all those technologies at the time relied on heavy scripting. In broad terms, when you spoke to them, software would translate your speech into text, analyze what you wanted, and create a response pulled from preapproved snippets. Those snippets could be charming, but they were also repetitive and simply boring—downright robotic. That was especially a challenge for a robot that was supposed to be social and family oriented. 

What has happened since, of course, is a revolution in how machines can generate language. Voice mode from any leading AI provider is now engaging and impressive, and multiple hardware startups are trying (and failing) to build products that take advantage of it. 

But that comes with a new risk: While scripted conversations can’t really go off the rails, ones generated by AI certainly can. Some popular AI toys have, for example, talked to kids about how to find matches and knives. 

OpenAI

Dactyl

A robot hand trained with simulations tries to model the unpredictability and variation of the real world.

By 2018, every leading robotics lab was trying to scrap the old scripted rules and train robots through trial and error. OpenAI tried to train its robotic hand, Dactyl, virtually—with digital models of the hand and of the palm-size cubes Dactyl was supposed to manipulate. The cubes had letters and numbers on their faces; the model might set a task like “Rotate the cube so the red side with the letter O faces upward.”

Here’s the problem: A robotic hand might get really good at doing this in its simulated world, but when you take that program and ask it to work on a real version in the real world, the slight differences between the two can cause things to go awry. Colors might be slightly different, or the deformable rubber in the robot’s fingertips could turn out to be stretchier than it was in simulation.

COURTESY OF OPENAI

The solution is called domain randomization. You essentially create millions of simulated worlds that all vary slightly and randomly from one another. In each one the friction might be less, or the lighting more harsh, or the colors darkened. Exposure to enough of this variation means the robots will be better able to manipulate the cube in the real world. The approach worked on Dactyl, and one year later it was able to use the same core techniques to do something harder: solving Rubik’s Cubes (though it worked only 60% of the time, and just 20% when the scrambles were particularly hard). 

Still, the limits of simulation mean that this technique plays a far smaller role today than it did in 2018. OpenAI shuttered its robotics effort in 2021 but has recently started the division up again—reportedly focusing on humanoids. 

Google DeepMind

RT-2

Training on images from across the internet helps robots translate language into action.

Around 2022, Google’s robotics team was up to some strange things. It spent 17 months handing people robot controllers and filming them doing everything from picking up bags of chips to opening jars. The team ended up cataloguing 700 different tasks.

The point was to build and test one of the first large-scale foundation models for robotics. As with large language models, the idea was to input lots of text, tokenize it into a format an algorithm could work with, and then generate an output. Google’s RT-1 received input about what the robot was looking at and how the many parts of the robotic arm were positioned; then it took an instruction and translated it into motor commands to move the robot. When it had seen tasks before, it carried out 97% of them successfully; it succeeded at 76% of the instructions it hadn’t seen before. 

COURTESY OF GOOGLE DEEPMIND

The second iteration, RT-2, came out the following year and went even further. Instead of training on data specific to robotics, it went broad: It trained on more general images from across the internet, like the vision-language models lots of researchers were working on at the time. That allowed the robot to interpret where certain objects were in the scene.

“All these other things were unlocked,” says Kanishka Rao, a roboticist at Google DeepMind who led work on both iterations. “We could do things now like ‘Put the Coke can near the picture of Taylor Swift.’” 

In 2025, Google DeepMind further fused the worlds of large language models and robotics, releasing a Gemini Robotics model with improved ability to understand commands in natural language. 

Covariant

RFM-1

An AI model that allows robotic arms to act like coworkers.

In 2017, before OpenAI shuttered its first robotics team, a group of its engineers spun out a project called Covariant, aiming to build not sci-fi humanoids but the most pragmatic of all robots: an arm that could pick up and move things in warehouses. After building a system based on foundation models similar to Google’s, Covariant deployed this platform in warehouses like those operated by Crate & Barrel and treated it as a data collection pipeline. 

By 2024, Covariant had released a robotics model, RFM-1, that you could interact with like a coworker. If you showed an arm many sleeves of tennis balls, for example, you could then instruct it to move each sleeve to a separate area. And the robot could respond—perhaps predicting that it wouldn’t be able to get a good grip on the item and then asking for advice on which particular suction cups it should use. 

This sort of thing had been done in experiments, but Covariant was launching it at significant scale. The company now had cameras and data collection machines in every customer location, feeding back even more data for the model to train on.

COURTESY OF COVARIANT

It wasn’t perfect. In a demo in March 2024 with an array of kitchen items, the robot struggled when it was asked to “return the banana” to its original location. It picked up a sponge, then an apple, then a host of other items before it finally accomplished the task. 

It “doesn’t understand the new concept” of retracing its steps, cofounder Peter Chen told me at the time. “But it’s a good example—it might not work well yet in the places where you don’t have good training data.”

Chen and fellow founder Pieter Abbeel were soon hired by Amazon, which is currently licensing Covariant’s robotics model (Amazon did not respond to questions about how it’s being used, but the company runs an estimated 1,300 warehouses in the US alone). 

Agility Robotics

Digit

Companies are putting this humanoid to the test in real-world settings.

The new investment dollars flowing to robotics startups are aimed largely at robots shaped not like lamps or arms but like people. Humanoid robots are supposed to be able to seamlessly enter the spaces and jobs where humans currently work, avoiding the need to retool assembly lines to accommodate new shapes such as giant arms. 

It’s easier said than done. In the rare cases where humanoids appear in real warehouses, they’re often confined to test zones and pilot programs. 

COURTESY OF AGILITY ROBOTICS

That said, Agility’s humanoid Digit appears to be doing some real work. The design—with exposed joints and a distinctly unhuman head—is driven more by function than by sci-fi aesthetics. Amazon, Toyota, and GXO (a logistics giant with customers like Apple and Nike) have all deployed it—making it one of the first examples of a humanoid robot that companies see as providing actual cost savings rather than novelty. Their Digits spend their days picking up, moving, and stacking shipping totes.

The current Digit is still a long way from the humanlike helper Silicon Valley is betting on, though. It can lift only 35 pounds, for example—and every time Agility makes Digit stronger, its battery gets heavier and it has to recharge more often. And standards organizations say humanoids need stricter safety rules than most industrial robots, because they’re designed to be mobile and spend time in proximity to people. 

But Digit shows that this revolution in robot training isn’t converging on a single method. Agility relies on simulation techniques like those OpenAI used to train its hand, and the company has worked with Google’s Gemini models to help its robots adapt to new environments. That’s where more than a decade of experiments have gotten the industry: Now it’s building big.