(Courtesy of Figure’s public live stream)

Figure’s robots just sorted packages for 200 hours straight

What started as a 10-hour human-vs-robot challenge turned into a continuous marathon shift spanning nine days of continuous work.

Jon Keegan

53m

Recently we’ve seen a lot of marathon-running, kickboxing, and moonwalking robots show off some impressive feats. But if bipedal humanoid robots are really going to be the multi-trillion dollar industry that Elon Musk predicts, they are going to need to do some economically valuable work.

Early this morning, the San Jose, California-based robotics company completed that kind of robotic flex: its Figure 03 robots worked a demonstration package sorting line for 200 hours straight. In a post on X, Figure founder and CEO Brett Adcock said that what started as a planned eight-hour challenge turned into a marathon sorting shift spanning 9 days.

We just wrapped what began as an 8-hour challenge - and it ran for 200 hours without a failure

Shoutout to the team for the hardcore engineering behind F.03 and the robust Helix models powering it pic.twitter.com/ir47UasJOn
— Brett Adcock (@adcock_brett) May 22, 2026

The live-streamed video showed a count of 249,560 packages sorted over 200 hours — that’s 8 days and 8 hours — the equivalent of 25 human 8-hour shifts. All with no bathroom or meal breaks.

The challenge started as a “Man vs. Machine head-to-head” challenge to see who could sort more packages over 10 hours – the Figure 03 robot, or Aime the human intern. Adcock noted:

“We’re following California labor laws, so the human gets both meal breaks and paid rest breaks during the shift.”

The task was to sort small packages — find the barcode, and place it face down on the conveyor belt. After 10 hours, the intern won, sorting 12,924 packages to the robot’s 12,735.

Adcock added his prediction to the post:

“This is the last time a human will ever win.”

Congrats to Aime!! He said his left forearm is basically broken 😂

Final scores:
→ F.03: 12,732 packages (2.83 seconds/package)
→ Aime: 12,924 packages (2.79 seconds/package)

This is the last time a human will ever win pic.twitter.com/CalDzPZz4d
— Brett Adcock (@adcock_brett) May 18, 2026

To be fair, this wasn’t one robot standing in place the entire time. A series of robots working in concert and communicating together swapped out when the working robot’s battery was low (which Adcock said was about every 3-4 hours). Then the active robot would step back and walk away to recharge, and the next robot would step in and pick up where the first robot left off.

Adcock said in a post that the task was completed autonomously by the team of robots, with “no humans in the loop.”

The company said it has finalized the design of the next iteration of the robot, Figure 04.

Rani Molla2h

Report: Uber considers full Delivery Hero takeover to take on DoorDash outside the US

Uber appears to be considering upping its competition with DoorDash outside the US, exploring a potential full takeover of Frankfurt-listed Delivery Hero, Bloomberg reports. Earlier this week the US-based ride-hailing service disclosed a 19.5% stake in the food delivery company, but now that could go higher.

The $11.8 billion German company could be particularly vulnerable to a takeover right now, with its CEO having recently stepped down following pressure from activist investors to sell off assets. A full acquisition would give Uber a massive foothold in over 60 countries to combat DoorDash’s European-focused Wolt unit.

Uber has been involved in a lot of deal-making of late, mostly in the autonomous vehicle space, where it now has more than 30 partnerships globally.

Uber extended its losses on the news and is currently down around 1.7%.

Uber Said to Study Full Takeover of European Rival Delivery Hero

Uber has been involved in a lot of deal-making of late, mostly in the autonomous vehicle space, where it now has more than 30 partnerships globally.

Uber extended its losses on the news and is currently down around 1.7%.

Rani Molla5h

Meta released a Reddit dupe. Reddit investors don’t like it.

Fresh on the heels of releasing a Snapchat dupe, which sent Snap down earlier this month, Meta seems to be meddling with Reddit, quietly releasing a Reddit-like Facebook app called Forum yesterday. After news of the “dedicated space built for deeper discussions, real answers and the communities you care about,” Reddit’s stock is down 4.5% today.

Last month, Reddit’s earnings report handily beat analysts’ expectations, but it continues to struggle with the perception that bigger tech companies — including Meta — investing heavily in AI will eat its lunch. The stock is down nearly 40% year-to-date.

Jon Keegan5/21/26

Report: OpenAI’s Q1 revenue was $5.7 billion, beating Anthropic

The neck-and-neck race between OpenAI and Anthropic as the AI companies barrel toward their expected IPOs this year is shaking out some internal numbers for would-be investors to ponder.

The Information is reporting that OpenAI’s first-quarter revenue was ~$5.7 billion, about $1 billion ahead of Anthropic’s revenue for the same period.

The Wall Street Journal recently reported that Anthropic is on course to more than double its first-quarter revenue of $4.8 billion to $10.9 billion in the second quarter. It is not known what OpenAI is projecting for Q2.

Recently, The New York Times reported that Anthropic’s current fundraising round seeking to raise between $30 billion and $50 billion comes with a valuation of up to $950 billion, putting it ahead of OpenAI’s latest reported valuation of $850 billion.

OpenAI Held $1 Billion Revenue Lead Over Anthropic in First Quarter

Rani Molla5/21/26

Alphabet’s Waymos are still getting caught in floods after recall

Waymo, the self-driving subsidiary of Alphabet, has paused operations in Atlanta after a new report of a vehicle driving into a flooded roadway and getting stuck, TechCrunch reports. The news comes just weeks after the company recalled its fleet of nearly 4,000 driverless cars to deal with a previous flood incident in San Antonio, where the service is also paused.

After that incident, Waymo instituted an “interim remedy” to make the vehicles “exclude additional operating conditions that present an elevated risk of encountering a flooded, higherspeed roadway,” but added that it was still “developing the final remedy for this recall.”

As we’ve noted, Waymo has mostly kept its rollout — now public in 11 cities — to more temperate climates, as severe weather poses more challenges to autonomous vehicles.

Waymo pauses Atlanta service as its robotaxis keep driving into floods | TechCrunch

As we’ve noted, Waymo has mostly kept its rollout — now public in 11 cities — to more temperate climates, as severe weather poses more challenges to autonomous vehicles.

Jon Keegan5/21/26

Report: Anthropic is in talks to use Microsoft’s custom AI chips

Anthropic is in talks to rent custom AI chips from Microsoft, according to a report from The Information, as the Claude coder’s scramble for compute continues.

During the first wave of the generative-AI boom, companies rushed to get their hands on Nvidia’s GPUs, as they were the only game in town if you wanted to build new models.

But as the role of inference has shifted to a top priority, with companies focusing on actually running models to make money, they’ve started shopping around, buying chips tailored for the task, and in some cases decided to make their own.

Additionally, Anthropic has become something of a victim of its own success at rolling out products that can be quickly adopted by enterprise clients. That rapid, wide-scale adoption has revealed significant compute constraints. Anthropic is now, effectively, looking for any and all compute capacity it can find, striking deals with CoreWeave, Amazon, Google and Broadcom, and even xAI.

Amazon and Google have both seen hot demand for their custom inference chips. But Microsoft is still trying to get its custom Maia chips into the mix, after encountering delays.

If Microsoft lands Anthropic as a customer for its Azure-based Maia computing services, it could open the door for other companies seeking another option for meeting the sky-high demand for AI inference, as agentic models gobble up trillions of tokens.