AW
Alexandr Wang
Co-founder (former CEO)
Previously Engineer at Quora
MIT (dropped out)
Scale AI provides data labeling, model evaluation, and AI infrastructure services to develop and improve AI applications for enterprises, governments, and frontier AI labs. The company also operates Scale Labs, a research division focused on AI safety, evaluation, alignment, and enterprise deployment.
A reasoned read from public sources. Each point links to its source.
Scale is a genuinely impressive business — $2B revenue, profitable data unit, real government traction, and a product moat in FDEs that took years to build. But the Meta deal is a strategic inflection point that cuts both ways: it provides capital and a committed customer, while simultaneously destroying the neutrality that made Scale trusted by every other major AI lab. The company is now effectively Meta's data arm in the eyes of the market, which is a very different business than the neutral infrastructure layer it was. The founder departure compounds this — Wang was Scale's most important sales asset. This is a company in genuine transition, not a stable compounder.1234
The data labeling market was valued at $6.5B in 2025 and is growing, while the broader agentic AI TAM is estimated at $40B in 2026. Scale commands roughly 35% of the dedicated data labeling/prep market. The company is actively pivoting toward agentic infrastructure and enterprise AI deployment, which represents a much larger addressable opportunity. Government contracts — including $300M+ in DoD deals — provide a meaningful diversification vector beyond commercial AI labs.5672
Scale's core product evolved from image annotation to a full-stack AI data engine: RLHF pipelines, LLM evaluation, red-teaming, and now the Scale GenAI Platform (SGP) for agentic workflows. The newly launched 'Dialect' product targets enterprise institutional knowledge capture — the gap between raw model capability and organizational trust. The forward-deployed engineer model (200+ FDEs embedded at OpenAI, Microsoft, Meta, DoD) is a genuine product moat that is hard to replicate quickly. However, the pivot from labeling vendor to 'agentic infrastructure layer' is still unproven at scale.89104
Alexandr Wang co-founded Scale at 19, grew it to $2B revenue, and was widely regarded as one of the most effective young CEOs in tech. However, in June 2025 he departed to join Meta as Chief AI Officer, leaving interim CEO Jason Droege (former Uber Eats, Axon) in charge. Co-founder Lucy Guo was fired two years after founding. The leadership transition is a significant uncertainty — Droege has operational pedigree but is unproven in the AI infrastructure domain, and Wang's departure removes the founder-market-fit that drove Scale's culture and customer relationships.11112131415
Revenue grew from $870M in 2024 to $2B in 2025 (~130% YoY), with the data business turning profitable and the applications business more than doubling in H2 2025. Scale raised $1B at a $14B valuation in May 2024, then Meta invested $14.3B for a 49% stake at a $29B valuation in June 2025. The company has 400+ enterprise clients and an average ACV of $2M. International public-sector revenue doubled in 2025 and is projected to double again in 2026. However, the Meta deal immediately triggered Google (~$200M annual contract), Microsoft, xAI, and OpenAI to sever ties, and Scale laid off 14% of staff.2161413153
Scale faces competition from Labelbox, Appen, Sama, Encord, Superb AI, and open-source tools like CVAT and Label Studio, plus emerging players like Thinking Machines Lab and Scematics. The neutrality crisis created by the Meta deal has made 'Scale AI alternatives' one of the most-searched topics in AI services for 2026, actively accelerating competitor growth. Commoditization pressure in core labeling is real — the market is pushing toward automated annotation and self-serve platforms. Scale's ~35% market share and FDE model are genuine advantages, but the customer trust damage from the Meta deal is a structural competitive wound.861718
Scale reached $2B revenue with ~130% YoY growth and its data business turned profitable — rare for a company at this stage. The $29B Meta-backed valuation provides a war chest and a committed anchor customer. The DoD and international public-sector business is growing rapidly and is structurally insulated from the commercial neutrality concerns. The FDE model and GenAI Platform represent a genuine moat in translating frontier model capability into enterprise-specific systems, and Dialect addresses a real unmet need in institutional knowledge capture.216144105
The Meta deal is an existential trust problem: Google (~$200M contract), Microsoft, xAI, and OpenAI all severed ties within days of the announcement, and Scale laid off 14% of staff. The founder who built every major customer relationship is now working for a direct competitor. The core labeling business faces commoditization as AI models improve at self-annotation. The pivot to agentic infrastructure and enterprise software puts Scale in direct competition with well-resourced incumbents (Salesforce, Microsoft, ServiceNow) on their home turf, without a proven track record in that category.1537
Scale would need to successfully replace lost commercial AI lab revenue (Google, Microsoft, OpenAI, xAI) with government and international public-sector contracts, which are growing but may not fully compensate. The GenAI Platform and Dialect would need to achieve meaningful enterprise adoption, validating the pivot from labeling vendor to AI infrastructure layer. Jason Droege would need to prove he can retain key FDE talent and rebuild customer trust without Wang. Finally, the core labeling business would need to avoid rapid commoditization long enough for the higher-margin software products to reach scale.5114109
This is a high-risk, high-ambiguity bet right now. Scale has real revenue, real product, and real government traction — but you'd be joining at a moment of maximum organizational uncertainty. The founder who set the culture and won the customers is gone. The company just lost its four largest commercial AI lab customers in a single week. A 14% layoff happened last summer. The interim CEO is unproven in this domain. If you're a senior engineer who wants to work on genuinely hard AI infrastructure problems with government-scale contracts, there's interesting work here — especially on the FDE, GenAI Platform, or defense side. But you should price in real execution risk: the company's strategic identity is unsettled, and the equity story depends on a pivot that hasn't been validated yet.114316
Scale AI provides data labeling, model evaluation, and AI infrastructure services to develop and improve AI applications for enterprises, governments, and frontier AI labs. The company also operates Scale Labs, a research division focused on AI safety, evaluation, alignment, and enterprise deployment.
Scale was founded in 2016 by Alexandr Wang and Lucy Guo through Y Combinator after they met working together at Quora. Wang, then 19, dropped out of MIT after his freshman year to build the company.
Axios interview with Scale AI CEO Jason Droege on pushing AI reliability and life after the Meta deal.
Scale launched Scale Labs, an expanded research division building on its SEAL lab and focused on model capabilities, post-training evaluation, enterprise deployment, and risk oversight.
Major AI lab customers including Google, OpenAI, and xAI paused or wound down work with Scale following Meta's investment, citing data confidentiality and competitive concerns.
Scale AI cut roughly 200 employees (14% of its ~1,400-person workforce) and 500 contractors, with interim CEO Droege citing over-scaling in the data-labeling business as key AI lab customers paused work.
Founder and CEO Alexandr Wang left Scale AI to lead Meta's superintelligence efforts as part of the Meta deal, while remaining on the Scale board.
Meta finalized a $14.3B investment in Scale AI valuing it at ~$29B, taking a 49% non-voting stake — Meta's largest investment since the WhatsApp acquisition.
Meta Platforms invested more than $14 billion to acquire a 49% non-voting stake in Scale AI at a $29 billion valuation; founder Alexandr Wang joins Meta.
Chief Strategy Officer Jason Droege was promoted to CEO after Alexandr Wang left to head AI efforts at Meta.
Scale AI announced partnership with Microsoft to develop AI-powered solutions for U.S. Indo-Pacific and European Command under the Defense Innovation Unit's Thunderforge program.
Scale AI reached a multimillion-dollar deal with the U.S. Department of Defense to develop the Thunderforge project, using AI to plan and execute movements of ships, planes, and other assets.
Co-founder (former CEO)
Previously Engineer at Quora
MIT (dropped out)
Co-founder (former)
Previously Product Designer at Quora
Co-founder; former CEO (now Director); Chief AI Officer at Meta
Previously Tech Lead at Quora
MIT (dropout)
Co-founded Scale AI in 2016 at age 19 after dropping out of MIT. Served as CEO until June 2025, when Meta invested $14.3B and recruited him to lead its superintelligence efforts. Remains on the Scale board.
Co-founder (departed 2018)
Previously Product Designer at Snapchat
Carnegie Mellon (dropout); Thiel Fellow
Co-founded Scale AI in 2016 with Alexandr Wang. A Thiel Fellowship recipient and Carnegie Mellon dropout who previously worked at Facebook, Quora, and Snapchat. Left Scale AI in 2018 due to differences in vision and later founded creator platform Passes.
$16B raised total
Petitioner on record
SCALE AI INC · SAN FRANCISCO, CA
FY 2025
See full record