What Hub.xyz does

Hub.xyz (legal entity Hub Data Inc., Delaware) is a real-world AI training data company founded in 2024 and operating as part of Y Combinator's P26 (Winter 2026) cohort. It supplies physical AI training datasets — egocentric video, image, video, and audio — to Fortune 50 enterprises and frontier AI labs developing robotics, autonomous systems, and embodied AI. The platform combines a global contributor network spanning 150+ countries with a verified SMB partner program across 10 industries (creative studios, pharmacies, medical clinics, hairdressers, cafés, restaurants, wood workshops, warehouses, production facilities, schools) to capture long-tail physical work data that is otherwise inaccessible.

The company offers four pre-built, professionally annotated off-the-shelf datasets (Egocentric, Image, Video, Audio) totaling roughly 54,000 hours of egocentric video, 2.4 million image frames, 70,000 video clips at 30fps, and 80,000 hours of multilingual audio at 48 kHz. Custom data collection projects are scoped by modality, volume, coverage, and complexity, with itemized quotes returned within hours and S3 delivery in roughly 36 hours. The end-to-end pipeline spans capture, professional editing (via 100+ Creative Studios), annotation (multi-object tracking, action segmentation, speaker diarization, 3D hand reconstruction), quality scoring, and delivery. Contributors onboard through platform.hub.xyz with OAuth via Google or Discord, undergo optional KYC through Didit, and are compensated in USDC on the Base network.

Hub.xyz runs a hybrid go-to-market: enterprise direct sales for dataset licensing and custom collection, paired with a product-led self-serve motion for contributor and SMB onboarding. Pricing is custom and not publicly disclosed. The team is sub-10 people led by co-founder and CEO Tim Sprecher, co-founder Armin Kiani, and Head of GTM Jeff Kanai. Marketing is concentrated on X (240K+ followers) and Discord (235K+ members), which together function as the primary contributor acquisition channels.

Name

Hub.xyz

Legal name

Hub Data Inc.

Website

https://hub.xyz

Company type

Private

Founded year

2024

Operating status

Operating

Headcount range

1–10 employees

Short description

Hub.xyz supplies Fortune 50 enterprises and frontier AI labs with annotated real-world training data across egocentric video, image, video, and audio modalities, sourced from a 150-country contributor network and 73+ verified SMBs across 10 industries.

Ownership category

akta.pro rank

Product category

AI Training Data

NAICS

Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518210), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518)

SIC

Services-Computer Programming, Data Processing, Etc. (7370), Services-Computer Processing & Data Preparation (7374)

akta.pro primary industry

Data Pipelines for GenAI (Curation, Filtering, Deduplication, Copyright) (HDAAACAJ)

akta.pro secondary industries

Dataset Discovery, Marketplaces & Licensing (HDAAALAE), Bias, Fairness & Representativeness Testing for Datasets (HDAAALAK)

Keywords

• AI training datasets
• Data annotation services
• Egocentric video capture
• Multimodal data licensing
• Custom data collection

Headquarters

HQ city

San Francisco

HQ country

United States

HQ region

North America

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Operations, Marketing or Sales, Infrastructure

Revenue model

1. Dataset Licensing: Hub licenses AI training datasets (egocentric video, image, video, audio) to enterprise clients including research labs and technology companies. Datasets include uploaded content with anonymized or aggregated demographic metadata.
2. Custom Data Collection Services: Bespoke long-tail data projects priced by modality, volume, coverage, and complexity. End-to-end pipeline from brief submission to S3 delivery with itemized pricing within hours.
3. Contributor Rewards: Contributors earn USDC rewards for completed tasks and approved uploads. Rewards processed as USDC payouts on Base network cryptocurrency wallet.

Pricing tiers

Go-to-market motion2 records

Distribution channels2 records

Marketing channels5 records

Core offering

Hub.xyz licenses AI training datasets across four modalities (egocentric video, image, video, audio) sourced from a global network of 150+ countries and 73+ verified SMB partners across 10 industries. Enterprise clients can purchase off-the-shelf datasets or commission Custom Data Collection projects delivered via S3 through an end-to-end pipeline that handles capture, annotation, and labeling. The company also operates a Contributor Platform (platform.hub.xyz) where individuals and SMBs complete data capture tasks in exchange for USDC payouts.

Product overview

Hub.xyz operates a data collection and annotation platform that provides AI training datasets for physical AI development. The core offering consists of four off-the-shelf dataset products organized by modality: Egocentric Video, Image, Video, and Audio datasets. These datasets are sourced from a global network of verified contributors (150+ countries) and SMB partners across 10 industries. The platform combines its pre-built datasets with a Custom Data Collection Service that enables bespoke, long-tail data projects. Contributors access the platform via the Contributor Platform for task completion and payouts, while enterprise clients can request specific data collection or license existing datasets. All offerings are structured around four data modalities (egocentric, image, video, audio) within a single unified data pipeline.

Differentiator

Problem solved

Functional benefit

Products and services

• Egocentric Dataset First-person POV video dataset capturing workplace and residential environments from diverse global locations. Includes IMU sensor data, RGB depth information, and rigged 3D hand reconstruction for training physical AI systems. Dataset comprises 54,000 hours of video across 84 environments.
• Image Dataset Diverse image collection sourced from the Hub contributor community and professionally edited through a network of 100+ Creative Studios. Includes raw and edited versions with annotations such as subject removal and bounding boxes. Contains 2.4M frames across 38 verticals at 1080p+ resolution.
• Video Dataset Pre-processed video clips with multi-object tracking annotations, action segmentation along timelines, and scene-level captions. Frame-accurate, pipeline-ready format. Contains 70,000 clips totaling 18,000 hours at 30fps native frame rate.
• Audio Dataset Multilingual field recording collection featuring speaker diarization, language and accent identification tags, and per-utterance emotion labels. Waveforms are aligned with transcripts for searchability. Contains 80,000 hours of audio across 47 languages at 48 kHz sample rate.
• Custom Data Collection Service Bespoke data collection service enabling enterprise clients to commission custom-scoped long-tail data projects. Priced by modality, volume, coverage, and complexity. End-to-end pipeline from brief submission to S3 delivery in hours rather than weeks, with first samples typically in S3 within 8.5 hours and full delivery within 36 hours.

Quantifiable outcome

• Custom data delivery in 36 hours from brief submission
• +2 more outcomes

Segments4 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration3 records

AI capability7 records

Feature5 records

Partnerships

Two partnerships are on record, tiered supporting and core.

• DiditsupportingTechnology or IntegrationDidit is a third-party KYC (Know Your Customer) provider used by Hub for identity verification of contributors. Contributors who choose to verify their identity have Didit collect government ID, selfie, and liveness check data. Hub only stores verification status, not identity documents.
• 100+ Creative StudioscoreStrategic or Co-development PartnerHub has a network of 100+ Creative Studios that professionally edit images sourced from the Hub community. These studios provide image annotation and editing services as part of the data pipeline.

Scale indicators10 records

Recent moves6 records

Expansion highlights6 records

Direct peers

• Defined.ai: Defined.ai operates a marketplace for AI training datasets including speech, image, and text across many languages. Its dataset-licensing model and enterprise customer base make it a direct comparable to Hub.xyz's dataset licensing and custom collection offerings.
• Scale AI: Scale AI is the largest AI training data platform, providing data labeling, annotation, and evaluation services across image, video, text, and audio modalities. It serves frontier AI labs and enterprises directly comparable to Hub.xyz's customer base, with established enterprise sales motion.
• Toloka AI: Toloka AI is a crowdsourced data labeling platform offering human-in-the-loop annotation across text, image, and audio for foundation models. Its global contributor network and AI lab customer base make it a direct peer to Hub.xyz's contributor-powered model.
• Appen: Appen is a publicly listed data annotation company offering image, video, speech, and text labeling via a global crowd workforce across 170+ countries. Its global contributor network and modality breadth directly mirror Hub.xyz's positioning and SMB-style crowd model.
• Roboflow: Roboflow specializes in computer vision training data, particularly for robotics and physical AI applications, offering image and video datasets plus annotation tools. Its robotics/embodied AI focus directly overlaps with Hub.xyz's egocentric and physical AI positioning.
• Sama: Sama provides high-quality training data annotation with a focus on computer vision and large-scale image/video labeling for AI applications. It targets the same enterprise AI/ML buyer set as Hub.xyz with comparable multi-modal offerings.
• Surge AI: Surge AI is a high-quality AI training data provider focused on expert-labeled datasets for foundation model developers. It competes with Hub.xyz in the same premium end of the data-for-AI market and shares Hub's focus on serving frontier AI labs.

Broad incumbents

• Labelbox: Labelbox is a data-centric AI platform providing annotation tooling and managed labeling services across multiple modalities. It is a broader incumbent with platform-software focus but overlaps with Hub.xyz in delivering labeled datasets for AI training.
• Snorkel AI: Snorkel AI provides a data-centric AI platform with programmatic labeling and dataset curation for enterprise ML. As a broader incumbent in data-centric AI, it overlaps with Hub.xyz's pipeline approach to building AI training datasets.

Emerging players

• Encord: Encord provides annotation and data management tooling for computer vision and multimodal AI, with growing managed services. As a newer entrant in video/image annotation, it overlaps with Hub.xyz's image and video modality products for AI/ML buyers.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat4 records

Key risks6 records

Key highlights6 records

Customer concentration

• Xx.com/hubxyz
• LinkedInlinkedin.com/company/104221584

Revenue estimate

Valuation estimate

Number of profiles

Profiles3 records

Funding overview

Funding rounds4 records

Investors2 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Hub.xyz