Robots Atlas>ROBOTS ATLAS
Robotics & Hardware

1X's Børnich sees a "hard takeoff" in 3 years, details NEO

1X's Børnich sees a "hard takeoff" in 3 years, details NEO

Bernt Børnich, CEO of 1X Technologies, put a concrete date on a robotics "hard takeoff" during an All-In podcast interview (July 29–30, 2026) — his bet is three years, with an upper bound of less than a decade. He also laid out how the NEO humanoid is meant to become an open platform, and where the company will get the data to train it.

Key takeaways

  • Børnich: a "hard takeoff" is his three-year bet, and surely less than a decade away.
  • Definition: robots autonomously build other robots, data centers, chip fabs, and refine raw materials.
  • NEO as an open platform — it runs competitors' models, including OpenAI and Anthropic.
  • The data strategy is a four-tier pyramid: from teleoperation to general internet video.
  • NEO price: $20,000 upfront or $499 a month. Consumer shipments in 2026.

What Børnich means by "hard takeoff"

A "hard takeoff" is the point where robots build other robots on their own — plus data centers, chip fabs, and refined raw materials. It is a loop in which production stops depending on human hands. Børnich does not frame it as distant speculation.

I am extremely sure that we're less than a decade away from hard takeoff.

Bernt Børnich, CEO and founder of 1X Technologies, on the All-In podcast.

It is a bold claim, but still a statement, not a dated roadmap. We treat it as the founder's position, not a confirmed schedule.

NEO as an open platform

NEO is not meant to be a closed product with a single brain. Børnich describes it as a developer platform that runs external models — including those from rivals like OpenAI and Anthropic. The robot works in three modes: full autonomy, shared autonomy with a human, and full teleoperation: A human remotely controlling the robot in real time..

The hardware is built to mimic a human — biomechanics, even skin friction. NEO's hands carry 22 degrees of freedom per side (44 total), which the company says goes beyond basic teleoperation fidelity. The robot stands 5'6" and weighs 66 lbs (about 30 kg), with a lifting capacity up to 154 lbs.

The data pyramid: from teleoperation to YouTube

The interview's real substance is the data strategy, which Børnich stacks into a four-tier pyramid — from the sparsest, best data to the most massive.

  1. Teleoperation — sparse but highest-quality data (top of the pyramid).
  2. Humans wearing the robot's sensors.
  3. Egocentric human video (first-person perspective).
  4. General internet video at YouTube scale — the base, which Børnich calls "ludicrously immense."

The key here is the cross-embodiment approach: the human body is the primary training target, and the robot's hardware mimics its mechanics, so human data maps onto the machine's motion.

Price and shipping

NEO is set to cost $20,000 upfront or $499 a month on subscription. A $200 deposit reserves an order. The company confirms consumer deliveries in 2026, though Børnich candidly notes the start will be "rough around the edges."

$20,000NEO price upfront — or $499 a month on subscription
It is worth separating facts from promises: price, specs, and "foundational autonomy" are company data, while NEO's ability to work in a home without remote supervision remains a goal for now. For hard tasks NEO doesn't know, a 1X expert can take over remotely during scheduled windows.

Why it matters

Børnich's claim shifts the humanoid debate from "whether" to "when," and a three-year horizon is an aggressive signal to the whole field. But the real content is in the data strategy — that, not the hardware, decides whether a humanoid learns real tasks. Opening NEO to rivals' models also inverts the usual market logic: 1X bets on being a platform, not just a hardware maker. It is a wager on where the value in this chain actually sits.

What's next?

  • NEO consumer deliveries are slated for 2026 — confirmed by the company, though without an exact start date.
  • The real test will be how much home work NEO does autonomously versus via a remote 1X expert.
  • The internet-video data model needs proving in practice — for now it is an interview claim, not published results.

Sources

Share this article