The AI Bubble and Web3: Are Tokenized AI, DePIN, and On-Chain Data Overpriced or Undervalued?

The AI bubble and Web3 is not a single story. Public-market AI leaders and AI infrastructure look richly valued, but they are supported by real revenue growth and substantial capital expenditure. Many tokenized AI, DePIN, and on-chain data projects, by contrast, still trade mostly on narrative momentum, with valuations that frequently lead fundamentals by a wide margin. At the same time, a smaller subset may be structurally undervalued if the market is missing durable cash flows, defensible data moats, or network effects.

This article breaks down where AI-related Web3 assets may be overpriced, where they could be undervalued, and how professionals and enterprises can evaluate risk using a fundamentals-first framework.

Macro Context: What People Mean by the "AI Bubble"

The AI boom has been intense and uneven. In 2025, AI-related companies drove a large share of U.S. equity gains, and sharp single-day drawdowns in leading chip stocks reinforced bubble narratives. Yet AI investment is not purely speculative. Mega-cap technology firms are committing very large AI budgets over the next several years, and AI data center demand is projected to grow at roughly 19% to 22% annually through 2030, driven by high-performance computing and storage needs.

Two competing interpretations can both be valid:

• Overextended growth, not a pure bubble: Much of AI infrastructure spending is tied to real adoption and revenues, with profitable firms funding capital expenditure from strong balance sheets.
• Speculative excess risk: Some economists and researchers argue that returns from scaling large language models may be diminishing, and that infrastructure buildouts could overshoot actual productivity gains.

For Web3, this matters because crypto markets often amplify major narratives. When AI optimism is strong, tokenized AI and DePIN can become high-beta expressions of the same theme, even when the underlying token projects have limited real usage.

Capital Flows: Why Hype Spills into Tokenized AI and DePIN

AI has absorbed a striking share of venture funding. In 2025, AI startups received around 41% of private-company funding, up sharply from 2021, with the majority of that capital concentrated in a small number of companies. Large technology firms simultaneously signaled aggressive increases in data center spending, and memory suppliers reported sharp profit growth and shortages linked to AI demand.

This combination creates a familiar dynamic in crypto markets:

• Massive, visible real-world demand for compute, storage, and data
• Smaller, more volatile token floats that can move quickly on narrative catalysts
• Valuation gaps between public equities priced on earnings and tokens priced on optionality

What Counts as Tokenized AI, DePIN, and On-Chain Data?

Tokenized AI

Tokens connected to AI services or infrastructure - such as decentralized inference, model marketplaces, AI agents, or tokenized access to model outputs. The key question is whether the token captures value from real usage or simply functions as a speculative wrapper.

DePIN (Decentralized Physical Infrastructure Networks)

Protocols that use token incentives to coordinate real-world resources such as GPU compute, storage, bandwidth, sensors, or energy. AI-focused DePIN typically targets compute and storage bottlenecks.

On-Chain Data

Data that is stored, referenced, or attested on-chain, or verifiably linked via oracles. In AI, this often includes provenance-attested datasets, licensing registries, labeling marketplaces, and verifiable inference outputs.

A Practical Bubble Framework for AI-Web3 Tokens

To assess whether the AI bubble and Web3 is real in a specific segment, a structured checklist is useful. A sound diagnostic examines economic strain, industry strain, revenue growth, valuation heat, and funding quality. Applied to tokenized AI, DePIN, and on-chain data, the picture often resembles a micro-bubble sitting on top of a genuine AI demand cycle.

• Economic strain: Web3 AI is not macro-significant yet. The strain shows up more in crypto drawdowns than in broader economic indicators.
• Industry strain: Many tokens imply large future adoption while current utilization remains modest - a classic mismatch.
• Revenue growth: Mixed. A few protocols show early fee traction; many report near-zero revenue.
• Valuation heat: Tokens often trade at extreme multiples of annualized revenue, or with no revenue at all.
• Funding quality: Ranges from established crypto backing to short-term retail flows and incentive farming.

Where Tokenized AI, DePIN, and On-Chain Data May Be Overpriced

1. Extreme Fully Diluted Valuation Relative to Protocol Revenue

A common overpricing pattern is a multi-billion fully diluted valuation paired with low single-digit millions, or less, of annualized protocol revenue. Even allowing for early-stage optionality, that gap is often wider than what public markets tolerate for AI leaders with established earnings.

2. Narrative-Driven Price Action

Many AI tokens rally around broad AI headlines - model releases, chip announcements, or "AI chain" marketing. When prices move more on global AI news than on project-specific usage metrics, bubble risk is elevated.

3. Low Utilization Versus Theoretical Capacity

In compute DePIN, the gap between "available GPUs" and "paid jobs executed" is significant. If a network cannot consistently match demand to supply, token incentives risk creating idle capacity and inflated expectations.

4. Weak Moats and Easy Forkability

Projects that function as thin orchestration layers over off-chain providers, without defensible data, strong reputation systems, or meaningful integration depth, are fragile at high valuations. If users can switch providers with minimal friction, long-term token value capture is uncertain.

5. Token Emissions That Dilute Fundamentals

High emissions can temporarily attract hardware and attention, but they can also suppress sustainable token value if organic demand does not materialize before rewards taper. Investors frequently underestimate how emissions shift effective valuation over time.

Where Some Projects Could Be Undervalued

Even in a frothy market, undervaluation can exist when a token trades like a narrative asset but behaves more like productive infrastructure.

1. DePIN with Real, Priced Resources and Repeat Demand

AI infrastructure demand is structurally strong, with data center demand projected to grow at roughly 19% to 22% annually through 2030. DePIN networks that deliver reliable compute or storage with verifiable utilization, competitive pricing, and consistent service quality can become meaningful suppliers. If markets price them as speculative AI beta rather than infrastructure with cash-flow potential, they may be underappreciated.

2. On-Chain Provenance for Compliance and Licensing

Enterprises are placing greater importance on training data provenance, copyright safety, and auditability. On-chain attestation and licensing registries can become more valuable as regulation and procurement processes require traceability. Tokens that power these registries may appear unremarkable today, which is precisely why they can be mispriced relative to long-term demand.

3. Transparent Fee Capture and Disciplined Token Economics

Some protocols are building clear fee models, limiting emissions, and publishing metrics that allow analysts to track real adoption. In a sector where many valuations ignore cash-flow-like signals, transparent value capture can be a source of structural undervaluation.

Real-World Use Cases: Where Value Is Created or Destroyed

DePIN for AI Compute and Storage

• Decentralized GPU marketplaces: Can monetize inference and training jobs and reduce bottlenecks, but must demonstrate reliability, scheduling, and compliance.
• Decentralized storage for datasets and model artifacts: Can charge storage and retrieval fees, but must compete on integrity, performance, and enterprise-grade guarantees.
• Edge compute and sensor networks: Can create unique data streams that are difficult to replicate, potentially forming defensible moats.

Tokenized AI Services and Model Marketplaces

• Model marketplaces: Value depends on whether the marketplace delivers aggregation, quality assurance, and distribution - not just listings.
• AI agents as on-chain actors: Promising for automation in DeFi and payments, but most deployments remain early and experimental.
• AI-enhanced DeFi: Must demonstrate measurable performance improvements and robust risk controls, not simply "AI inside" branding.

On-Chain Data and Verifiable Inference

• Data provenance and licensing registries: Can support compliant training pipelines and creator compensation models.
• Labeling and validation markets: Require strong sybil resistance and dispute resolution, or incentives get misdirected.
• Verifiable inference and zkML: Could become critical for trustworthy on-chain agents and high-stakes workflows where auditability is essential.

How Professionals and Enterprises Can Evaluate Pricing in AI-Web3

For developers, analysts, and enterprise teams, the goal is to separate infrastructure-like networks from narrative-driven tokens.

A Due Diligence Checklist

1. Utilization: What percentage of compute, storage, or data capacity is paid for and used regularly?
2. Revenue quality: Are fees paid by external users, or mostly recycled incentives?
3. Unit economics: Are providers profitable at market prices without emissions support?
4. Token value capture: Do fees accrue to token holders, stakers, or a treasury in a measurable way?
5. Moat: Is there a defensible dataset, reputation layer, compliance posture, or integration advantage?
6. Emissions and dilution: How does circulating supply evolve over 24 to 48 months?
7. Compliance and abuse prevention: How does the network prevent illicit compute use and meet enterprise requirements?

Professionals looking to build expertise in this area can explore Blockchain Council training across AI, Web3, DeFi, and blockchain security. Relevant programmes include the Certified Artificial Intelligence (AI) Expert, Certified Blockchain Developer, Certified Web3 Professional, and Certified DeFi Expert certifications.

Conclusion: Bubble Signals Are Real, but Mispricing Cuts Both Ways

The AI bubble and Web3 debate is best resolved through segmentation. AI infrastructure in public markets is expensive, but backed by revenue, capital expenditure, and measurable macro impact. In token markets, many AI and DePIN assets display recognizable bubble traits: extreme valuation multiples, narrative-driven rallies, low utilization, and unclear moats.

That said, the picture is not entirely frothy. DePIN networks with real utilization and durable unit economics, along with on-chain data and provenance protocols positioned for compliance-driven demand, can be undervalued when markets treat them as short-term narrative trades. The practical opportunity lies less in calling the entire sector a bubble and more in identifying which protocols are building infrastructure that enterprises will actually pay for.