Figure AI Reports Rapid Growth for Index, Surpassing 69,000 Weekly Active Users

Key Takeaways

Show all

• Figure AI has grown its Index crowdsourcing platform to 69,943 weekly active users as of early September 2026, up from roughly 44,000 at launch.
• The platform’s ingestion rate has accelerated to 35 minutes of physical video uploaded every second, up from 30 minutes per second reported in August.
• The consumer data pipeline directly feeds Figure's proprietary Helix foundation model, aimed at solving zero-shot general-purpose manipulation.
• The surge in crowdsourced data follows Figure's multi-billion-dollar compute pact with Nscale, underscoring the company’s bet on data scale over mechanical iteration.

Two weeks after formally unveiling Index, Figure AI founder and CEO Brett Adcock posted updated adoption metrics, signaling rapid user growth for the company's video data collection initiative.

According to data shared by Adcock, Index recorded 69,943 weekly active users for the week ending September 4, 2026. Along with the increase in active participants, data ingestion expanded: contributors are now uploading 35 minutes of video every second across the globe, an increase from the 30 minutes per second cited during the platform's public debut in late August.

Scaling Up the Ingestion Funnel

Index, which operated quietly under the internal moniker Project Go-Big for four months prior to its public launch, incentivizes smartphone owners to record first-person footage of everyday physical manipulation tasks. The crowdsourced library spans domestic chores like making beds and folding laundry to commercial activities in retail and logistics.

A weekly active user chart published by Adcock highlights the platform’s ramp-up curve:

• May 1, 2026: 99 weekly active users
• June 5, 2026: 2,668 weekly active users
• July 10, 2026: 8,656 weekly active users
• August 7, 2026: 31,055 weekly active users
• August 21, 2026: 44,436 weekly active users
• September 4, 2026: 69,943 weekly active users

At a velocity of 35 minutes of footage uploaded per second, the pipeline is processing roughly 5.7 years of physical human activity every 24 hours.

Fueling the Helix Pipeline

Figure positions this continuous stream of human demonstration as the cornerstone of its Helix foundation model architecture. While traditional robotics approaches lean heavily on teleoperation or synthetic physics engines, Figure is attempting to replicate the scaling laws of large language models by feeding raw real-world visual interactions into high-capacity multimodal models.

Raw uploads pass through an automated five-stage ingestion pipeline consisting of technical filtering, fraud screening, embedding-based deduplication, task rebalancing, and hierarchical text annotation. The resulting structured dataset is designed to help Figure’s humanoids achieve zero-shot generalization across novel objects and unstructured environments.

The acceleration in data intake also puts Figure's massive infrastructure plans into context. Just last week, the company announced a multi-billion-dollar compute agreement with Nscale to deploy up to 100,000 next-generation NVIDIA GPUs starting in late 2027.

Capital and Data as the Primary Moats

Adcock has repeatedly argued that hardware manufacturing is no longer the central roadblock for the industry. Having built more than 1,000 units of the Figure 03 and deployed machines into enterprise pilots with BMW and Catalyst Brands, leadership maintains that onboard intelligence remains the defining constraint.

With $15 million already disbursed to contributors and over $1 billion committed toward data acquisition and compute over the next year, Figure is treating real-world data collection as a capital-intensive scale problem.

Open questions remain regarding whether monocular smartphone video can provide the fine-grained force dynamics and tactile feedback required for intricate physical assembly. Nonetheless, the early growth trajectory of Index indicates that finding willing human contributors is not the bottleneck in Figure’s equation.