DePIN in 2026: Who Actually Buys Crowdsourced Network Data?

Six months ago, I started building QuackNet, a decentralized physical infrastructure network (DePIN) for crowdsourced network intelligence. The whitepaper was clean: users scan WiFi, cellular, BLE, and IP data on their phones. We aggregate it. We sell it to enterprises. We pay users in NTI tokens. Everyone wins.

Then I had to answer the question that kills most DePIN projects: Who actually buys this?

This isn't a theoretical post. This is what I learned building a real DePIN platform and talking to actual customers—telcos, CDN operators, regulators, and smart city platforms. The gap between DePIN whitepaper vision and market reality is brutal. Let me walk you through the four buyer segments that actually exist, the pricing models that work, and the uncomfortable truth about token economics.

The DePIN Landscape in 2026

DePIN (Decentralized Physical Infrastructure Network) has matured since 2024. Helium proved the concept works at scale—but also proved the massive challenges. Here's the honest state of play:

• Helium (mobile, $MOBILE): 500K+ hotspots, proven coverage. But the April 2023 independent operator transition was chaotic. The network effect is real but fragile. Revenue? Token value-based, not from actual mobile services revenue.
• Hivemapper (dashcam-based street mapping): 200K+ contributors, 20%+ of US roads covered. But licensing to Google Maps/Apple Maps is the actual business—tokens are secondary incentive.
• DIMO (vehicle data): 100K+ cars connected, but still pre-revenue. Enterprise pilots (insurance, fleet) are underway but not scaled.
• Silencio (noise mapping): Smaller but focused. EU regulators actually mandate noise monitoring, so there's real regulatory demand.
• QuackNet (network intelligence): We're at 200 beta testers. 932K+ French cell towers imported. Real APIs with 3 paying pilots.

What they all share: crowdsourced physical-world data + token incentives. What they all struggle with: the cold-start problem (need data to sell revenue, need revenue to incentivize data).

The 4 Real Buyer Segments for Network Intelligence

Stop thinking "enterprise" as one blob. Network data has four distinct buyers with completely different use cases and willingness to pay.

1. Telcos (Orange, SFR, Bouygues, Free) — Coverage Verification

Here's the dirty secret: telcos have terrible coverage data. Their maps are based on:

• Tower propagation models (theoretical, 2+ years old)
• Their own network probes (biased—they own the towers)
• OpenSignal/Tutela data (but that's also crowdsourced and they can see it too)

What they don't have: actual user experience. Dead zones. Indoor penetration gaps. Handover failures at highway off-ramps. That's where crowdsourced data wins.

What they pay for: API access to coverage heatmaps (H3 hex resolution), quality metrics per tower, signal statistics, handover event logs. Annually: €50K–€200K per country depending on geographic coverage density and historical data depth.

What they don't care about: Token prices. They're not holders. They want accurate, fresh, auditable data.

2. CDN/Cloud Providers (Cloudflare, Akamai, Google, AWS) — Last-Mile Latency Mapping

CDNs make money by delivering content fast. To do that, they need to know: where should edge nodes go? Where is latency actually bad?

Traditional data (speed test services) doesn't cut it because:

• Speed tests measure arbitrary locations, not real user traffic patterns
• Latency varies wildly by ISP, peering agreements, backbone routing
• Crowdsourced backbone pings (traceroute hops to major IXPs) reveal actual user-to-IXP paths

When we deployed backbone pings to 8 major European IXPs (France-IX, AMS-IX, DE-CIX, LINX, SwissIX), we found something interesting: Parisian users on SFR had 25ms-to-France-IX latency, but Orange users had 65ms—because of routing contracts, not geography. That's worth millions if you're deciding where to invest in edge capacity.

What they pay for: Per-geography latency distributions, hop-by-hop AS path analysis, ISP peering relationship inference, edge node deployment optimization. Annually: €100K–€500K depending on region coverage and realtime data freshness.

3. Smart City & IoT Platforms — Infrastructure Density Mapping

Cities want to know: where are WiFi dead zones? Where do residents have poor cellular? Where's the BLE/IoT density? This data drives:

• Municipal WiFi rollout planning (e.g., Paris deploying free WiFi in underserved arrondissements)
• 5G rollout prioritization (Bouygues vs SFR vs Orange coverage gaps by district)
• Real estate valuation (connectivity = property value; this is a major factor now post-2022)
• IoT infrastructure planning (if your city is deploying smart parking/trash/weather sensors, you need to know BLE/LoRaWAN coverage)

The killer insight: BLE density data is proprietary to Apple/Google. Cities can't see where iBeacons, AirTags, and Tile devices cluster. That's a new data type we can monetize—we know BLE device density per H3 hex. That tells cities "this neighborhood has 40K smart home devices, this one has 8K."

What they pay for: Coverage heatmaps (WiFi/cellular per district), BLE density overlays, network quality segments (hospital/school/park quality requirements), competitive carrier maps. Annually: €30K–€150K depending on city size.

4. Regulators (ARCEP, Ofcom, FCC, BNetzA) — Independent Verification

This is the most underrated buyer. Regulators can't trust telco self-reporting. They have no independent data. When they grant spectrum licenses or approve consolidation (e.g., 3G+4G merger), they need independent coverage proof.

ARCEP now mandates "independent third-party coverage verification" for:

• Universal Service Obligation compliance (specific coverage %targets)
• License renewal conditions
• Merger/acquisition reviews
• Subsidy programs (€2B+ for rural broadband rollout in France)

Crowdsourced data at sufficient scale (10K+ scans per region) becomes regulatory-grade evidence. It's auditable, reproducible, hard to game, and independent.

What they pay for: Annual coverage reports per carrier/region, statistical confidence intervals, equipment validation, historical trending. Annually: €200K–€1M depending on jurisdiction and coverage depth required.

The Pricing Problem That Everyone Gets Wrong

Most DePIN projects price like SaaS (seats, API calls, tiers). That's wrong. Network data is a commodity with a cost-of-goods-sold problem: you have to pay users to generate it.

Let's do the math. France is ~550K km². To get "regulatory grade" coverage (10+ scans per 174m hex = H3 res-9), you need:

• ~7 million hexes at res-9
• 70+ million scan data points (assuming 10 scans per hex as minimum)
• At €0.10 per scan (NTI token value), that's €7M in incentive spending

How do you fund that? By selling data at prices high enough to justify the incentive spending while leaving profit margin. Here's what actually works:

Compare this to OpenSignal ($50M+ raised) and Tutela (acquired for $100M+ by Opensignal). They command premium pricing because they have 10+ years of data history, mobile app distribution, and brand recognition. A new entrant needs to undercut or offer something unique.

Token Economics vs Real Revenue: The Uncomfortable Truth

Here's what I realized halfway through building QuackNet: tokens are a user acquisition tool, not a business model.

Most DePIN projects (Helium, Hivemapper, DIMO) rely on token appreciation to fund operations. The math:

• Raise venture: $20M
• Launch token: $MOBILE, $HONEY, $DIMO
• Pay it to users: spend €500K/month in tokens to incentivize data
• Hope token appreciation covers the burn
• Eventually: the token becomes worthless (inflation + no revenue), and the project dies

Helium's $MOBILE crashed 90% because they could never generate real revenue to back token value. Hivemapper works because they have a clear path to licensing revenue (Google Maps, Apple Maps). DIMO is still unproven.

QuackNet's approach is different:

1. Tokens are incentive-only. NTI tokens are paid to users proportional to data quality + geographic scarcity. No token speculation required.
2. Real revenue comes first. Enterprise API subscriptions fund the operation. Token value is derived from platform success, not speculation.
3. The token is a cost of goods. Like Stripe's payout cost or AWS's instance cost, we budget NTI distribution as a known expense (€2–€5 per 1000 scans).

This is less sexy than "we have a token!" but it's the only model that doesn't collapse when crypto crashes.

Comparison: Where Each DePIN Project Stands (March 2026)

Project	Data Type	Scale	Buyer Segment	Revenue Model	Token Dependency
Helium	Cellular coverage	500K hotspots	None (failed)	Token ($MOBILE)	100% (dead)
Hivemapper	Street imagery	200K dashcams	Google, Apple, TomTom	Data licensing	50% (secondary)
DIMO	Vehicle data	100K cars	Insurance, fleet ops (pilot)	Licensing (pre-revenue)	80% (primary incentive)
Silencio	Noise levels	50K sensors	EU regulators, cities	Regulatory reporting	20% (marginal)
QuackNet	Network intelligence	200+ beta (target 10K yr1)	Telcos, CDN, regulators	Enterprise API subscriptions	25% (incentive cost)

What Building QuackNet Actually Taught Me

1. The Cold Start Problem Is Real, and It's Not Solved by Tokens

You need critical mass of data before anyone buys. Collecting that data costs money upfront (user incentives). You can't sell data you don't have yet. This is a chicken-egg problem that VC funding solves, but it's brutal if you're bootstrapping.

Solution: Focus on one geographic region first (e.g., Île-de-France) rather than global. Get to regulatory-grade density in one region, sell to ARCEP + 1-2 telcos, fund expansion to next region.

2. Data Freshness Matters More Than Historical Depth

Enterprises pay for current coverage maps, not historical trending. A 3-month-old coverage map is worth 10% of a current one. This means continuous scanning (not one-time). It's why subscription revenue is more reliable than one-time licensing.

3. Regulatory Buyers Are Slow, But They Pay 5x More

Getting an ARCEP contract takes 9+ months of validation. But when you land it, it's a multi-year, multi-million Euro commitment. Compare that to CDN pilots (3-6 months, €50-200K annually). Regulatory business takes longer to build but has much higher lifetime value.

4. The Real Differentiation Is Data Quality, Not Coverage

Everyone says "we have data." What they mean: "we have some data." Quality beats quantity:

• 10K high-quality scans (rich metadata, verified GPS, signal logs) beat 100K low-quality ones (SSID only, coarse location)
• Fraud detection matters at scale. If 10% of your scans are bot/fake, your data is worthless to telcos.
• Our fraud detection (4-check engine: velocity, frequency, plausibility, signal) catches 70% of obvious bot farms. Only ML will get to 95%+.

5. You Need All Data Types to Win

Single-axis data (WiFi only, cellular only, BLE only) has limited buyer value. Telcos want WiFi + cellular together. CDNs want cellular + traceroute + latency. Regulators want all of it cross-validated.

We collect 5 data types (GPS, WiFi, Cellular, IP, BLE) + enrichment (altitude, barometric pressure, neighbor cells, traceroute hops, backbone pings, device metadata). That combination is hard to replicate.

Where This Is Going (2026–2027)

Here's what I expect:

• Regulatory buyers go mainstream: By 2027, ARCEP, Ofcom, and FCC all mandate independent coverage verification. This is a €500M+ TAM globally.
• Data moats emerge: First-mover advantage in a region (France, UK, Germany) means you have 18+ months of data history. That's worth 3-5x more than fresh data alone.
• AI/ML fraud detection replaces rules: Rule-based fraud detection maxes out at 70% accuracy. ML will push it to 95%+, unlocking 10x better data quality.
• Token economics separate from user incentives: Projects that still rely on token appreciation will quietly move to fiat-based incentive models. Tokens become cosmetic.
• Android gold data is the next battleground: iOS limits cellular data access. Android gives raw CellID. Whoever scales Android in Europe first wins the tower inference game.

The DePIN vision—crowdsourced infrastructure with token incentives—is real. But it's not magic. The winners will be the ones who figured out that tokens are a tool, not the business model. The real value is in the data itself.

Final Take

Building QuackNet forced me to question every assumption in the DePIN playbook. The honest answer to "who buys this?" is: four specific buyer segments, each with 10-100x different pricing expectations, and only the ones with regulatory or latency-critical use cases will pay real money for scale.

If you're building a DePIN project, stop focusing on token price and start talking to regulators, CDNs, and telcos. That's where the money is. The token will follow.