Scale AI is not a large language model company. It does not build chatbots or compete with OpenAI, Anthropic, or DeepSeek. Instead, Scale AI is the infrastructure backbone that trains frontier AI models—the "picks and shovels" of the AI revolution. The company labels training data, teaches AI systems through reinforcement learning from human feedback (RLHF), and operates classified government contracts for the US Department of Defense. On June 10, 2025, Meta Platforms made a stunning move: investing $14.3 billion for a 49% non-voting stake in Scale AI, valuing the company at $29 billion post-money and signaling that data quality infrastructure may be more critical to winning AI than raw model development.

What makes this story investable is that Scale AI—founded in 2016 by then-teenager Alexandr Wang—has transformed from a Y Combinator-backed startup into a data engine essential to every frontier AI company, and is now backed by Meta, Amazon, and institutional investors controlling $15.9 billion in cumulative funding across nine rounds. Wang, now 29 years old, became the world's youngest self-made billionaire in 2025. Yet Scale AI has not gone public. For most investors, direct ownership remains unavailable. However, understanding Scale AI's role in the AI infrastructure stack, the Meta partnership's implications, and pathways to indirect exposure is crucial for anyone building an AI-focused investment thesis.

This article maps Scale AI's founding, explosive growth trajectory, Meta's landmark investment decision, government contracts worth $99-100 million, competing business models, the realistic IPO timeline (or lack thereof), and investment routes available to both accredited and retail investors. We examine whether Scale AI represents defensive infrastructure that benefits from AI proliferation regardless of which models win, or concentrated risk dependent on Meta's continued commitment.

Also in this series: How to Invest in OpenAI | How to Invest in Anthropic | How to Invest in DeepSeek

Scale AI's Founding: A Teenager's Y Combinator Bet on Data Labeling

Scale AI was founded in 2016 by Alexandr Wang and Lucy Guo, both of whom met while working at Quora. Wang was 19 years old at the time. Rather than studying computer science full-time at a traditional university, Wang enrolled in Y Combinator's Summer 2016 batch with Guo as co-founder and product designer. Their insight was straightforward but profound: machine learning models require massive amounts of labeled training data, but manual data labeling was expensive, slow, and error-prone. Scale AI would build software to orchestrate labeling at scale—combining human expertise with algorithmic workflows to produce clean, high-quality training data.

At the time, this was not a glamorous market. Computer vision and natural language processing were advancing, but the bottleneck was not model architecture—it was data. Scale AI positioned itself to solve this problem by becoming a data labeling platform that served computer vision, autonomous vehicle, and NLP companies. Y Combinator backed them with a typical $150,000 seed investment, and Wang and Guo began approaching enterprises.

By 2018-2019, Scale AI had become the go-to data partner for autonomous vehicle companies and AI labs. When Peter Thiel's Founders Fund invested $100 million in Scale AI in August 2019, the company achieved unicorn status (valuation exceeding $1 billion) in just three years. This was a watershed moment: a teenager's data labeling startup was now valued at over $1 billion, signaling that infrastructure companies serving AI development could command venture capital at scale.

Explosive Growth and Funding Trajectory: From $1B to $29B in Six Years

Scale AI's funding rounds and valuations tell a story of sustained capital appetite and accelerating growth:

Funding Round	Date	Valuation	Lead Investors
Seed → Series A	2016-2018	$200M-$500M	Y Combinator, Accel, various
Series B (Founders Fund)	August 2019	$1.0B (unicorn)	Founders Fund ($100M)
Series C	July 2021	$7.0B	Greenoaks, Dragoneer, Tiger Global
Series D	2023	~$7.3B	Various growth investors
Series E	May 2024	$13.8B	Accel ($1B), Amazon, Nvidia, Meta
Meta Strategic Investment	June 2025	$29.0B	Meta ($14.3B for 49% stake)

The trajectory is remarkable: from $1 billion (2019) to $7 billion (2021) to $13.8 billion (May 2024) to $29 billion (June 2025). This 29x growth in valuation over six years reflects both Scale AI's operating success and broader market recognition that AI infrastructure commands premium multiples. The May 2024 Series E round was already notable for its size ($1 billion from Accel alone) and investor composition (Amazon and Nvidia entering as investors). But Meta's June 2025 investment redefined Scale AI's strategic position entirely.

The Meta Inflection: $14.3B for Data Superiority and Llama Training

On June 10, 2025, Meta Platforms announced a strategic investment in Scale AI that stunned the AI industry. The deal structure: Meta would invest $14.3 billion (reported as $14.8 billion in some sources) for a 49% non-voting stake in Scale AI. The company would be valued at approximately $29 billion post-money. Critically, this was not a controlling stake—Meta received minority ownership with no voting power—but it represented the largest single investment ever made by a Big Tech company into a private AI infrastructure startup.

Why would Meta make such a massive bet? The strategic rationale was clear: Meta CEO Mark Zuckerberg had invested billions in AI infrastructure and hiring but was frustrated with Llama's competitive positioning relative to OpenAI's GPT-4 and other frontier models. The bottleneck was not compute or model architecture. It was training data quality. Llama, despite being an impressive open-source model, was perceived as less capable than closed-source competitors in part because of data quality issues. By acquiring a 49% stake in Scale AI, Meta secured privileged access to Scale's data labeling, RLHF, and model evaluation infrastructure—technology that could directly improve Llama's training and competitive performance.

The deal came with a leadership change. Alexandr Wang, Scale AI's CEO and largest shareholder, accepted a position at Meta as Chief AI Officer and head of Meta's "Superintelligence Labs"—the division tasked with building AI systems that approach or exceed human-level intelligence. Wang stepped back from day-to-day Scale AI operations, with Jason Droege, Scale AI's Chief Strategy Officer (formerly at Uber), becoming the new CEO. This arrangement allowed Meta to benefit from Wang's AI expertise while maintaining his ownership stake in Scale AI.

Revenue, Growth, and the "Data Engine" Business Model

Scale AI's business model is fundamentally different from OpenAI or Anthropic. Rather than developing proprietary models, Scale AI monetizes through:

• Data labeling services: Customers pay per-token or per-task for labeled training data.
• RLHF (Reinforcement Learning from Human Feedback): Scale AI contracts to collect human feedback used to train and align LLMs.
• Model evaluation: Testing and benchmarking third-party AI models against Scale's evaluation frameworks.
• Scale GenAI Platform: A proprietary platform enabling enterprises to modify, test, and deploy generative AI models using their own classified or proprietary data.
• Donovan Platform: A defense-focused AI system for Pentagon and DoD use, built on Scale's infrastructure.

Scale AI's revenue growth has been extraordinary:

• 2023: Approximately $760 million annualized run rate (ARR)
• 2024: $870 million in revenue (calendar year); $1.5 billion ARR by December 2024
• 2025 projection: $2 billion+ in revenue (representing 130% year-over-year growth)

This growth profile—nearly doubling revenue year-over-year—is exceptional for a company already generating nearly $2 billion in annual revenue. The driver is clear: as frontier AI companies proliferate, demand for high-quality training data and RLHF infrastructure increases. Every major LLM—from OpenAI's GPT models to Google's Gemini to Meta's Llama—requires Scale's services at some point in the training pipeline.

The "Picks and Shovels" Thesis: Why Infrastructure Trumps Models

Scale AI exemplifies the classical "picks and shovels" investment thesis. During the California Gold Rush, the merchants who sold picks and shovels to miners often became wealthier than the miners who struck gold. Similarly, in the AI boom, infrastructure companies that enable model development across the industry can generate more stable, diversified revenue than any single model company.

Consider: whether OpenAI's GPT-5, Google's Gemini 2.0, or a hypothetical competitor emerges as the dominant model, all of them require high-quality labeled data and RLHF infrastructure. Scale AI's customers are every frontier AI lab, every enterprise implementing AI, and every government agency developing AI capabilities. This customer diversification is a moat, providing stability across different market conditions and investment scenarios. Scale AI doesn't depend on any single model winning; it profits as long as AI development accelerates globally.

The META investment partially contradicts this thesis—Meta now owns 49% of Scale AI, creating concentration risk. However, even with Meta as a major shareholder, Scale AI continues to work with Amazon, Microsoft, Google, and other customers outside Meta's ecosystem. The company is contractually obligated to serve these clients at arm's length.

Products: Data Labeling, RLHF, Donovan, and Scale GenAI

Core Data Labeling and RLHF

Scale AI's flagship offering is a data engine that combines AI-assisted pre-labeling with expert human review. Customers define labeling tasks (e.g., "identify objects in images," "rate conversational quality," "detect safety violations"), and Scale AI orchestrates a distributed workforce to label data at scale. The company manages both a curated in-house team and an external contractor network, enabling rapid scaling without losing quality control.

Scale's RLHF (Reinforcement Learning from Human Feedback) service is particularly critical for LLM development. As OpenAI demonstrated with GPT-3 and ChatGPT, RLHF—collecting human preferences for model outputs and using those preferences to fine-tune models—is essential for creating models that are both capable and aligned with human values. Scale AI's RLHF infrastructure is used by multiple frontier labs to train reasoning models, instruction-following models, and safety-aligned variants.

Donovan: The Classified Defense Platform

In May 2023, Scale AI signed a historic contract with the US Army's XVIII Airborne Corps, becoming the first AI company to deploy an LLM on a classified military network. This system, known as Donovan, enables intelligence analysts and military planners to query classified documents, generate insights, and support decision-making without transferring sensitive data to cloud providers.

Donovan represents a $99-100 million contract award (reported in 2025) and signals the Pentagon's willingness to outsource AI development to private companies like Scale AI. The platform is now offered to other DoD agencies and intelligence services, creating a dedicated revenue stream from government customers.

Scale GenAI Platform

Scale's GenAI Platform is an enterprise offering enabling organizations to deploy, modify, and test generative AI models using their own proprietary or classified data. This is distinct from OpenAI's API or ChatGPT's interface—it's a full development environment for organizations building custom AI applications without outsourcing data to third parties.

Government and Defense Contracts: $99-100 Million DoD Agreements

Scale AI's government contracts represent a significant and growing revenue stream. Beyond the Donovan contract with the Army, Scale AI has been awarded multiple DoD and intelligence community contracts:

• Donovan Program (Army): ~$99 million contract announced in 2024-2025, providing classified LLM infrastructure for Army R&D and intelligence operations.
• Pentagon Flagship Program (Thunderforge): Scale AI is a key partner in the DoD's Thunderforge program, developing AI agents for military planning and operations. This partnership includes collaborations with Microsoft, Anduril, and other strategic technology partners.
• Data Engine Services: Scale provides data preparation and model evaluation services to the Pentagon for testing and validating LLMs for defense applications.

These government contracts are valuable not just for revenue but for validation and market positioning. US defense and intelligence agencies only work with companies they trust with classified information. Scale AI's Donovan deployment proved the company's capability in this domain, likely opening doors to additional government customers and contracts.

The Competition: Labelbox, Appen, Surge AI, and Internal Teams

Scale AI is not alone in the data labeling and RLHF market. Competitors include:

Labelbox: A platform-first competitor that provides annotation infrastructure and a contractor network. Labelbox focuses on enabling enterprises to build internal labeling workflows rather than relying entirely on Labelbox's managed services. The company is well-funded and competes on flexibility and customer control. Labelbox is also pursuing an IPO, which could provide a public-market benchmark for Scale AI's valuation.

Appen: A publicly traded company (APEN on ASX) that operates a large-scale crowdsourced labeling workforce (1+ million workers globally). Appen competes on scale and cost, offering low-price labeling services. However, Appen's quality is often perceived as lower than Scale's premium offering, and the company has faced challenges competing with Scale AI for enterprise customers.

Surge AI: A venture-backed startup that has differentiated itself through a "elite labelers" model—hiring highly educated, specialized contractors rather than crowdsourced workers. Surge charges 10x more than Scale AI per task but claims superior quality. Surge reportedly achieved $1.2 billion in revenue without any venture capital, suggesting a highly profitable business model. However, Surge's premium pricing limits market reach compared to Scale's scale-oriented approach.

In-house labeling teams: Major AI companies (OpenAI, Google, Meta, Anthropic) have built internal data labeling teams, reducing dependence on external vendors. This represents structural competition that Scale must navigate. However, even with internal teams, frontier labs typically still use external partners like Scale for specialized tasks, spikes in demand, or government-cleared services.

Scale AI's competitive moats include: brand recognition among frontier labs, government contract relationships, integration with leading AI platforms, and the breadth of services (data labeling + RLHF + evaluation + GenAI + Donovan). Competitors either specialize in one area or attempt to compete on cost; Scale competes on comprehensive capability.

Alexandr Wang: From Y Combinator to Youngest Self-Made Billionaire

Alexandr Wang's rise is unusual in tech. Founding a company at 19 is remarkable; building that company to a $29 billion valuation by age 29 is extraordinary. Wang has become the world's youngest self-made billionaire—a title that undersells his actual wealth, as his 49%+ ownership stake in Scale AI (post-Meta investment) is worth approximately $14+ billion.

In January 2025, Wang attended President Donald Trump's second inauguration alongside other tech founders and made a public statement that "America must win the AI war"—positioning himself as a stakeholder in US AI competitiveness and national security. This political positioning, combined with his government contracts and appointment as Meta's Chief AI Officer, makes Wang not just an investor-grade founder but a political and strategic figure shaping US AI policy.

Wang's continued leadership (now at Meta rather than Scale) is important for investor assessment. As an executive at Meta with a $14+ billion stake in Scale AI, Wang is incentivized to make decisions that benefit both companies. Whether this creates conflicts of interest or synergies is an open question for investors.

Post-Meta Challenges and Competitive Headwinds

Meta's investment in Scale AI created an immediate and significant problem: OpenAI, Google, and other major customers became uncomfortable with the partnership. If Meta owned 49% of Scale AI, could Meta access or influence Scale AI's data or insights about competitors' models? This trust issue triggered mass customer departures in mid-2025.

Shortly after Meta's investment announcement, OpenAI, Google, and xAI all announced they would reduce or discontinue use of Scale AI's services. In July 2025, Scale AI laid off approximately 200 employees in its data labeling business—a direct consequence of customer defections. This was a stark reminder that even a $29 billion company with Meta backing faces structural competitive risks when customer trust erodes.

The challenge for Scale AI going forward: maintain customer relationships with AI labs that are now competitors of Meta, while simultaneously maximizing value from Meta's partnership. This is a delicate balance. Scale AI must convince OpenAI, Google, and others that the company remains an independent vendor despite Meta's majority stake. Contractual separations, data firewalls, and governance commitments are essential but may not fully assuage customer concerns.

IPO Timeline and Public Stock Access

Scale AI has not announced an IPO and has not publicly endorsed a timeline for going public. However, the company's scale ($29 billion valuation, $2 billion+ annual revenue) and investor composition (Meta, Amazon, Accel, Founders Fund) make an eventual IPO likely within 3-5 years.

Why an IPO is probable:

• Scale and maturity: $2 billion+ annual revenue with positive unit economics suggests readiness for public markets.
• Investor expectations: Accel, Tiger Global, Founders Fund, and other early-stage venture investors typically exit through IPOs or acquisitions within 8-10 years of initial investment.
• Employee liquidity: As Scale AI grows, employee retention requires stock liquidity options (secondary markets or IPO).
• Strategic independence: An IPO would signal to customers that Scale AI is independent and not controlled by Meta, potentially mitigating customer defection risks.

Why an IPO is uncertain:

• Meta's stake: Meta owns 49% of Scale AI but without voting rights. If Meta wanted to exercise influence over an IPO timeline, it could, though legally it would require Scale's board approval.
• Customer concentration risk: An IPO filing would require disclosure of customer concentrations. If OpenAI, Google, or other major customers have reduced orders (as reported in 2025), this would be disclosed and could depress IPO valuations.
• Government contracts complexity: The Donovan contract and other classified DoD work may complicate SEC disclosure and foreign ownership restrictions for IPO investors.

Most likely IPO timeline: 2026-2027, assuming Scale AI stabilizes customer relationships and demonstrates that the Meta partnership does not compromise vendor independence. A 2028-2029 IPO is also plausible if customer defection takes longer to overcome.

Public Stock Exposure Through Meta

For investors unable or unwilling to purchase Scale AI pre-IPO shares, the most direct public-market exposure is through Meta Platforms (META). Meta's $14.3 billion investment represents a material asset on Meta's balance sheet. While Meta doesn't account for this as a "investment" in traditional accounting (it's more of a strategic acquisition structure), Meta does own a 49% stake in a $29 billion company.

How to think about META as a Scale AI proxy:

• Upside case: If Scale AI's valuation doubles to $58 billion within 3-5 years (driven by AI model training becoming more competitive and government contracts expanding), Meta's stake would be worth ~$28 billion, representing significant value creation.
• Downside case: If Scale AI's valuation compresses due to commoditization of data labeling or customer defections, Meta's investment could depreciate. This would likely be a non-cash accounting charge but could impact Meta's balance sheet optics.
• Liquidity uncertainty: Meta's stake is likely illiquid until a Scale AI IPO or acquisition. If Meta wanted to exit before an IPO, the only option would be a secondary sale to another buyer, which could be difficult.

Buying META stock for Scale AI exposure is indirect and diluted—Meta's stock is driven primarily by advertising revenue, metaverse investments, and AI costs, not by a single portfolio company investment. However, as Meta's Superintelligence Labs ambitions accelerate, Scale AI becomes increasingly central to Meta's strategy, potentially becoming a larger component of investor thesis over time.

Pre-IPO Investment Markets: EquityZen, Forge, Hiive, and Secondary Platforms

For accredited investors willing to take on illiquidity and valuation risk, secondary markets exist to purchase Scale AI shares before a public IPO:

• EquityZen: Operates a secondary market for private company shares. EquityZen may list Scale AI shares from existing shareholders willing to sell. Transaction volumes are typically thin, and pricing may not reflect true fair value.
• Forge Global: Another secondary marketplace for pre-IPO shares. Forge connects buyers with existing shareholders and handles the legal and compliance mechanics of the transaction.
• Hiive: A newer entrant in pre-IPO trading, Hiive allows accredited investors to buy and sell shares in private companies. Scale AI shares have traded on Hiive at valuations ranging from $7-12 per share (implied company values of $22-38 billion), though volumes are low and bid-ask spreads are wide.
• Nasdaq Private Market (NPM): Nasdaq operates a private company trading platform where Scale AI shares may be listed and traded among accredited investors.

Important caveats: These secondary markets are largely illiquid compared to public stock exchanges. Bid-ask spreads can be 10-20% or wider. There is no guarantee you can exit a position quickly at the price you want. Additionally, secondary market valuations may not align with future IPO pricing—a share trading at $10 on Hiive might IPO at $20 or $5, depending on market conditions and company performance.

Investment Risks: Concentration, Commoditization, and Customer Defection

Meta Concentration Risk

Meta owns 49% of Scale AI—nearly half the company. While Meta has no voting rights, its economic stake is massive and creates several risks:

• Customer defection: As noted, OpenAI and Google have already reduced Scale AI usage due to Meta ownership concerns. If this trend accelerates, Scale AI's revenue growth could stall.
• Strategic pivot risk: Meta could pressure Scale AI management to prioritize Meta's AI needs over other customers, further alienating competitors.
• Regulatory scrutiny: Meta's stake could attract FTC or international regulatory attention if authorities believe Meta is using Scale AI to gain competitive advantages or control the data infrastructure market.

Commoditization of Data Labeling

Data labeling is becoming increasingly automated. AI pre-labeling (using smaller models to suggest labels, which humans then correct) is standard practice and reduces human effort. As pre-labeling improves, the cost of data labeling could fall significantly, compressing Scale AI's margins.

Additionally, if frontier AI labs build proprietary in-house labeling capabilities (as OpenAI, Google, and Meta have), demand for external vendors may decline. Scale AI would need to differentiate on specialized services (RLHF, classified government work, industry-specific expertise) rather than commodity labeling.

LLM Model Efficiency and Reduced Training Demand

DeepSeek's emergence in January 2025 demonstrated that frontier AI can be trained with vastly less compute and potentially fewer high-quality labeled examples through architectural efficiency. If frontier labs require less training data to build capable models, demand for Scale AI's data services could decline.

Conversely, more capable models might require even higher-quality, more specialized training data. The direction is uncertain, but improved model efficiency is a potential headwind for a company whose business is rooted in supplying training data.

Government Contract Concentration

Scale AI's government contracts ($99-100 million Donovan award) are valuable, but they are concentrated with the US DoD and intelligence agencies. If US government AI spending is redirected, or if the Donovan contract is not renewed after the initial award term, Scale AI would face revenue loss.

Additionally, the Trump administration's recent focus on AI regulation and export controls could shift government AI strategy in ways that benefit or harm Scale AI.

The "Picks and Shovels" Thesis: Supporting or Contradicting Evidence

Scale AI's business model—providing infrastructure that all AI companies depend on—should make it resilient to any specific model or application winning. However, several factors complicate this thesis:

Supporting evidence: Despite OpenAI and Google reducing orders post-Meta investment, Scale AI still works with Amazon, Microsoft, and enterprise customers. The company's $2 billion+ annual revenue and 130%+ year-over-year growth suggest strong underlying demand for data infrastructure that extends beyond any single customer.

Contradicting evidence: Meta's 49% stake means Scale AI is no longer truly agnostic infrastructure. The company is partially controlled by one of the largest AI companies, creating conflicts of interest that may limit its appeal as a neutral vendor. Additionally, frontier labs increasingly build in-house capabilities, reducing dependence on external vendors.

Overall assessment: Scale AI resembles infrastructure, but less because it's neutral and more because data is a fundamental input to all AI development. As long as AI companies require labeled data and RLHF, Scale AI has customers. Whether Scale AI can maintain independence from Meta and remain a trusted vendor to OpenAI, Google, and others will determine whether it fully realizes the "picks and shovels" premium or becomes more commodity-like.

Investment Verdict: Is Scale AI a Buy?

For different investor profiles, the verdict differs:

Accredited investors seeking pre-IPO exposure: Scale AI is an intriguing but risky bet. The company is valuable, growing rapidly, and backed by Meta and major VCs. However, the Meta stake creates structural challenges (customer defection), and the IPO timeline is uncertain. If you purchase Scale AI shares on secondary markets at a 5-10% discount to recent valuations and can tolerate 3-5 year illiquidity, the risk-reward may be acceptable. Suggested allocation: <5% of portfolio.

Retail investors unable to access pre-IPO markets: Wait for the IPO (likely 2026-2027). Scale AI will eventually go public, and retail investors will have equal access to IPO shares at the same terms as institutional investors. Trying to purchase pre-IPO shares through secondary markets as a retail investor exposes you to 1) illiquidity, 2) opaque valuations, 3) unfavorable bid-ask spreads, and 4) regulatory restrictions. It's not worth the friction.

Investors wanting AI infrastructure exposure now: Consider META as an indirect proxy. Meta's $14.3 billion Scale AI investment is a material long-term bet on data infrastructure. Buying META gives you exposure to Scale AI (49% ownership) plus Meta's core advertising and AI business. However, this is diluted—META stock is driven by many factors beyond Scale AI.

Long-term AI portfolio builders: Scale AI is a "must own" once it goes public. A data labeling and RLHF company valued at $29 billion with government contracts and frontier lab partnerships is an essential component of AI infrastructure exposure. Upon IPO, allocate 1-3% of portfolio to Scale AI as part of a broader AI infrastructure thesis (which might also include semiconductor, data center, and power companies).

Conclusion: The Hidden Engine of Frontier AI

Scale AI is not the flashy frontier of AI—it's not building models, not competing with OpenAI or Anthropic, not racing for AGI. Instead, it's the infrastructure that makes frontier AI possible: the labelers, the annotators, the RLHF engineers, the classified government platforms. Alexandr Wang understood at 19 that data is the bottleneck, not compute or talent. That insight, plus nine years of execution and a $14.3 billion validation from Meta, has made Scale AI a $29 billion data engine indispensable to the entire AI industry.

The Meta investment changes Scale AI's risk profile. It eliminates capital constraints but creates customer concentration risk and governance complexity. For investors, the key questions are: Will Scale AI maintain vendor independence and regain OpenAI/Google business? Can the company defend against commoditization and internal capabilities at frontier labs? Will the Donovan government contract become a major revenue stream or remain a niche offering?

These questions will be answered over the next 18-36 months. A successful IPO in 2026-2027 with customer diversification and government contract growth would validate Scale AI as essential AI infrastructure. A failed IPO, accelerated customer defections, or margin compression would suggest the market was overvaluing data labeling infrastructure.

For now, Scale AI remains an intriguing but high-risk private company. Accredited investors can take positions on secondary markets. Retail investors should wait for the IPO. And Meta shareholders should recognize that a significant portion of their equity is now linked to Scale AI's continued success in an increasingly complex and competitive AI infrastructure market.

"Scale AI is proof that in the AI era, the companies that provide the picks and shovels can be more valuable than the gold miners themselves—as long as they remain trusted by all miners, not just one."

Also in this series: How to Invest in OpenAI | How to Invest in Anthropic | How to Invest in DeepSeek

Disclaimer: This article is for informational and educational purposes only and does not constitute financial, legal, or investment advice. Scale AI is a private company with a complex ownership structure involving Meta Platforms. Investment in Scale AI pre-IPO shares carries substantial risks including illiquidity, valuation uncertainty, customer concentration risk, regulatory uncertainty regarding government contracts, and strategic dependence on Meta's continued support. Past revenue growth and Meta's investment do not guarantee future performance. The IPO timeline is speculative and dependent on market conditions, Scale AI's ability to resolve customer defection issues, and regulatory approval. Retail investors should not attempt to purchase pre-IPO shares on secondary markets without understanding liquidity and valuation risks. Investors should conduct thorough due diligence, consult with qualified financial and legal advisors, and carefully assess their risk tolerance before making any investment decisions. Frontier Ledger does not endorse any specific investment or provide personalized financial advice. Markets are volatile, and all equity investments carry risk of capital loss. This article reflects information available as of March 2026; circumstances and Scale AI's business conditions may change rapidly.