Humanoid Shipments Quadrupled in Six Months. Booster Robotics Claims a Top-Three Spot.

The humanoid robot industry keeps breaking its own records. Global shipments reached nearly 25,000 units in the first half of 2026, according to fresh IDC figures, up 432.1 percent from the same stretch a year earlier. The market passed $740 million in value. And IDC now expects annual shipments to clear 750,000 units by 2030, a forecast it raised by roughly half on the strength of this data.

The more interesting question is who is moving all these machines. Agibot, which shipped more than 8,600 units for roughly 35 percent of global volume, took the top slot from Unitree, per IDC data reported in early October. And on October 14, a far younger name forced its way into the conversation: Booster Robotics, founded in 2023, announced it ranked among the world's top three humanoid vendors by shipment volume.

That claim comes from the company's own press release, so apply the usual discount. But the reasoning behind it is worth examining, because Booster's story is the clearest version yet of where the humanoid business is actually headed: not toward the most cinematic robot, but toward the most usable platform.

The numbers behind the surge

IDC's tracker is the most bullish of the major reads, but the direction is consistent. China remains the industry's center of gravity: the domestic market shipped more than 19,000 units in the first half, up 426.3 percent year over year, about 78 percent of the global total. A second research firm, Smart Analytics Global, put the global tally lower, at roughly 19,100 units and 272 percent growth, but agreed that Chinese manufacturers built more than 97 percent of them. However you count it, the concentration is extreme.

The machines are starting to show up in real work, too. Agibot crossed 20,000 cumulative robots produced in September. More than 300 of its robots are now operating at Chimelong Spaceship Park in Hengqin, handling visitor guidance, retail services, and hospitality inside a high-traffic venue. Another 100 Agibot X2 humanoids are doing customer greeting and in-store wayfinding across ASD stores. These are still small fleets, but they are doing continuous commercial work, not demo routines.

Why developers are Booster's distribution strategy

Here is what makes Booster's claim worth a second look. Rather than chasing headline industrial contracts, the company has built its business around the people who will actually program these machines. Its pitch is a full stack: humanoid hardware, its own operating system (Booster OS), and Booster Studio, an integrated development environment built for embodied AI. The idea is that a lab or startup can start building embodied AI applications without designing hardware or infrastructure from scratch.

The contest is shifting from who builds the most impressive robot to who builds the most useful robot platform.

The receipts are unusual for a three-year-old company. Booster says its robots are in use across more than 40 countries and regions, spread across universities, research labs, competition teams, and commercial customers. At RoboCup 2026, 38 of 59 competing teams ran on Booster platforms, and Booster-based teams swept the championships in all three divisions: Small, Middle, and Large. In July, the company hosted Season 1 of its Booster Champion simulation competition and registered more than 1,000 teams worldwide in two weeks.

That same month Booster launched the T2, its flagship humanoid, aimed at real-world execution with heavy payload capacity, high hardware reliability, full-body teleoperation, and open connectivity. The company says the T2 is gaining adoption among top global clients, including in the United States.

The industry is grading on a new curve

IDC's report makes a point that echoes Booster's strategy directly: competition is shifting away from isolated hardware specifications toward broader embodied AI capabilities, things like environmental understanding, task planning, manipulation, and continuous learning inside real workflows. The machines matter less than the loop between deployment and improvement.

This is the familiar platform story from computing history. A robot that ships with an operating system, development tools, and an active community accrues advantages a standalone machine cannot: more experiments, more edge cases discovered, more software written by people who do not work for the manufacturer. Booster's stated vision, making humanoid robots "as simple, reliable, and practical as computers," is almost comically literal about it.

What to watch, and what to doubt

Skepticism is still warranted. Nobody outside IDC's client base has published the full vendor ranking, and "top three by shipment volume" is the most flattering frame available to a young hardware company: units moved, not revenue, not margins, not utilization. The entire disclosed revenue base of two major Chinese robot makers totaled just $235 million across 6,579 units in 2025, a reminder that shipping robots and making money from them are different businesses.

Then there is the geopolitical concentration. When one country builds more than nine out of ten humanoids on the planet, buyers elsewhere face supply-chain questions that have nothing to do with torque curves. Western champions like Figure AI, which cleared a $39 billion post-money valuation, are betting that software, integration, and local production can beat hardware volume. Whether that bet pays off will decide whether the 750,000-unit market of 2030 is a global industry or a Chinese one with guests.

For now, the signal from IDC's half-year data is unambiguous: the humanoid industry has crossed from demos into deployment. The question has changed from whether humanoids will ship at scale to whose platform the software gets written on. Booster's answer, bet on the developers, is at least a coherent one. And this week, for the first time, it has third-party numbers behind it.