Open Edge — Reinventing the ISP

Now

ISPs now: connectivity companies

Internet Service Providers laid the subsea cables, pulled the fiber and raised the towers, and absorbed the capital risk of the physical Internet. The value settled elsewhere: with platforms that hold the user relationship and the data it produces, while the network that carries everything is priced as a commodity.

01

Centralization was the winning architecture of the last era

Applications became data-centric, and data gravitates to where it is stored: the datacenter. Distributed machines could not coordinate persistent state, search and recommendation, so computation followed the data into central hubs, today’s hyperscale platforms.

Click for in depth

02

ISPs run the Internet’s one irreplaceable layer, and capture none of its value

Every online interaction passes through an ISP network built on decades of capital risk: subsea cables, fiber, towers. Yet connectivity sells as commodity transit, and ISPs are structurally excluded from the economics of the traffic they carry.

Click for in depth

03

The Internet runs on ads or subscriptions; most people fit neither

Advertising finances products by capturing and measuring attention; subscriptions recover cost only from frequent, committed users. The subscription-saturated, the never-payers and the occasional user fall outside both: their usage is booked at zero, with no mechanism to charge for it.

Click for in depth

In depth · 01

The founding protocols of the Internet were stateless and peer-to-peer by design. SMTP and HTTP moved mail and pages between independent machines and kept no record of their own, and the network’s early shape matched its architecture. The workload then changed. Users came to expect persistent accounts, real-time search and algorithmically curated feeds, and each of these requires state: data that must be stored somewhere physical, kept consistent, and queried at interactive speed.

During the formative years of the web, coordinating that much state across a loose, heterogeneous network was an intractable computer science problem. Storing and processing it in one place was not. Data gravitates to where it is stored, computation gravitates to the data, and monetization follows both. Routing traffic through central computational hubs, where aggregation, indexing and curation could be executed efficiently, was the most efficient design available, and the hubs built for it grew into today’s hyperscale datacenters.

None of this was monopolistic intent; centralization won because it worked, and for two decades it was the correct engineering answer. Its side effect was structural: the industry’s roles were assigned, with services and their economics concentrating in the hubs while the network beneath was left to carry traffic.

In depth · 02

Internet Service Providers are the only universal counterparties on the Internet: every online interaction, whatever its application or origin, passes through the network of the user’s ISP. No other class of actor holds this position. It rests on physical facts, wires into households, spectrum and cable landings, and on a direct, recurring billing relationship with nearly every connected household and business.

The economics of the position do not match its structure. ISPs financed the physical Internet: subsea cables, terrestrial fiber and cellular towers, capital committed up front and recovered over decades. The services running over this infrastructure settle their economics elsewhere, while connectivity itself is sold as undifferentiated transit, compared and bought on bandwidth and price. An ISP can invest, consolidate or discount, but within the current arrangement it cannot participate in the value of the traffic it carries.

The exclusion is historically understandable. Participating in service economics has so far meant becoming a platform: operating services, holding user data, competing with one’s own customers. ISPs lacked a neutral mechanism for taking part without abandoning their role as carriers. The position, the physics and the billing relationship remain in place; what has been missing is the instrument.

In depth · 03

Two monetization models finance effectively every consumer service on the Internet. Advertising sells the user’s attention: the service is free at the point of use, earns per impression, and is therefore engineered to maximize engagement and to measure behavior, because measured behavior is what advertisers buy. Subscriptions sell commitment: a flat recurring fee that recovers its cost only from users whose consumption is regular enough to clear the price.

A large share of real demand falls outside both boundaries. Households already carrying several subscriptions refuse another commitment at any price. Others never subscribe at all. The occasional user, one match, one film, one query a month, sits below every subscription’s break-even and is invisible to the model. Per-use payment does not rescue these cases: attaching a checkout to each interaction triggers the pain of paying and loss aversion, and the interaction simply does not happen.

The result is a market that undercounts its own demand. Usage with genuine value to both sides is booked at zero because no mechanism exists to charge for it, and services optimize for the two constituencies they can bill, heavy users and advertisers, rather than for use itself.

Act I

ISPs becoming clearing companies

The Edge Settlement Protocol federates ISPs into a shared metering and clearing layer. Users keep paying the one bill they already pay; applications never invoice, token-gate or collect cards. Every request is metered at the edge and settled automatically from an aggregated clearing pool.

01

Existing connectivity bill is used to settle all Internet service usage

The connectivity bill becomes a passport to the whole catalog. Like a city multipass covering every gym and pool, users pay a modest blended premium for commitment-free access to everything they actually touch — and subscription fatigue disappears.

Click for in depth

02

Clearing Internet service usage is a new revenue stream for ISPs

Every metered request clears through the ISP. The operator advances settlement, validates usage and takes a defined share of cleared volume — a predictable new revenue line derived from the settlement work itself, not from the bandwidth it sells.

Click for in depth

03

Metered approach to Internet service usage liberates users from subscriptions and advertisements

Because every service settles on the single connectivity bill, users escape the pile of per-app subscriptions — and because usage is paid for directly, the everyday services people rely on no longer need to run on advertising. One metered relationship replaces both.

Click for in depth

In depth · 01

For the household, no new commercial relationship is created; instead, the existing contract with its Internet service provider is used. The only change is the scope of the bill: the single recurring payment that already covers connectivity now also covers everything accessed through it, settled according to measured usage.

Metering itself is not new to households. Electricity, telephony, and other utilities have been billed according to measured usage for decades. However, each of these services has its own independent commercial relationship with the household and can therefore be metered and billed separately.

ISP contract becomes a form of universal Internet account. When consuming Internet services under that "universal account", payment is not attached to individual requests, usage is recorded at the point of interaction and settled afterward through the ISP bill; for most common services, no purchase decision occurs at the moment of use; a search, a minute of video or a query to a model adds to a meter rather than prompting a transaction. The overall goal is shifting household's perception of Internet service consumption into the package thinking — household assesses its Internet as a whole, across a billing period, rather than pricing each service in isolation — the unit of evaluation becomes the aggregate relationship, not the individual services.

In depth · 02

The ISP sits between the application delivering a service and the user consuming it, and guarantees that the two settle. Concretely, the operator advances funds from a pre-committed pool, validates the metered record of each unit of usage, and bears the cost of reconciliation and chargebacks. This intermediary role is compensated by fees levied on cleared volume. For the operator it is a revenue line tied to settlement, independent of the bandwidth it sells.

That revenue only appears where applications actually settle through the operator. Bilateral clearing between each application and each ISP does not scale and would require a separate commercial and technical arrangement for every pair, which applications would not undertake, and a single operator’s subscriber base is rarely enough to justify the work. The clearing role therefore needs a shared settlement surface that many operators present as one.

Federating ISPs using a neutral protocol brings such an integration surface. Because operators clear through one protocol rather than each on its own terms, an application integrates against the protocol only once and reaches the users of every federated ISP. The protocol stands as one counterparty to each side: a single settlement interface facing the application, a single clearing pool facing the operator.

In depth · 03

Subscription pricing requires a forward commitment: the household must judge, before trying a service, whether future use will justify a recurring fee. The structure filters adoption toward services used daily and against the occasional or unfamiliar. Each adopted service adds a standing contract to be tracked and eventually cancelled, and revenue under the model concentrates in payments that continue unnoticed.

Advertising-funded services carry no fee but are not costless: the user supplies personal data and attention, and product design is oriented toward the advertiser’s objectives. The model persisted because no other mechanism could finance services that are free at the point of use. Metered settlement supplies that mechanism. Usage is measured at the point of interaction and settled through the existing connectivity bill.

A single account then provides access to the entire catalog. Access held available regardless of use has a price of its own, known as option value, and households pay it knowingly: one projecting twelve dollars of ordinary usage will accept a bill of fifteen, with the premium pricing the option rather than expected consumption. Services with infrequent use or small audiences, unviable under subscription or advertising models, become accessible because no advance purchase decision is required.

Act II

ISPs becoming data companies

In the legacy web, one company’s crawler reads the web, one company’s cars photograph the streets, and everyone’s behavior is siphoned in secret to sell ads. ESP inverts the model: data is collected in the open, aggregated under strict privacy guarantees, and published into a commons no one owns.

01

ISPs make up the largest, widest data ingest in the world, paving the way for the largest datasets ever created

Cars record the roads they travel, home robots learn from daily tasks, and everyday Internet use carries economic signal. Today that value evaporates or accrues to device manufacturers. The federated ISP is the natural steward: it collects only what the household chooses to share, and puts it to work on the household’s behalf.

Click for in depth

02

Ingested and aggregated data become public commons datasets, not private data silos

What the network ingests doesn’t disappear into a proprietary vault. Anonymized, aggregated datasets are published to an open commons anyone can build on — AI pipelines, researchers, competitors — dissolving the private data moat that hyperscalers guard today.

Click for in depth

03

Data aggregation work is a new revenue stream for ISPs

Producing and refreshing those datasets is real work — and the protocol pays for it. Buyers fund the collection, not exclusive ownership, so aggregation becomes a new, recurring revenue line for ISPs that flows back as a visible discount on the household’s bill.

Click for in depth

In depth · 01

Households are about to produce more valuable data than most companies do. A car with cameras records the streets it drives through. Every task a household robot performs, from vacuuming today to housework tomorrow, is a recorded example of real physical work. Everyday browsing adds a record of what people look for and choose. Even the broadband line itself has a use, since a crawler running from a residential connection sees the ordinary web — the local versions, the pages served to real visitors — that datacenter crawlers get blocked from or shown substitutes for.

No single company is in a position to gather these streams together. The car maker gets its own cars, the robot vendor its own robots, and even the largest platform sees only what happens inside its own products. An ISP already has a paid relationship with the household, a legal presence in the same country, and a wire into every one of those devices. Thousands of ISPs acting through a common protocol can therefore collect across all of it, and can do it in the open: data is gathered when a buyer commissions it, each operator handles only its own subscribers, personal detail is removed before anything is published, and the household gets a share back via its regular monthly bill. Whether the car’s cameras contribute at all, and to which network, stays the owner’s decision rather than the manufacturer’s.

What comes out is hard to obtain by any other means: street imagery refreshed by daily commutes, manipulation data from household robots at population scale, a continuously updated index of the web, behavioral aggregates covering entire countries. Building any one of these requires being present in millions of homes at once, and that is the position ISPs already occupy.

In depth · 02

The defining rule of the data market is that the buyer never takes possession of the data. A purchase funds a round of collection, and the output — anonymized and aggregated until no individual can be recovered — is published openly for anyone to read. What a buyer pays for is the data existing and being current: the slice it needs is often missing or stale until someone funds its collection. Rivals reading the same dataset afterward takes nothing away from that, and a commissioned refresh still costs far less than running a private collection network — the only other way to get the data at all.

Because the dataset is open and excludes no one, no single buyer has to fund all of it. Different organizations pay to refresh the parts they care about — a logistics firm keeping road and street imagery current, a retailer funding collection of shopping behavior, a government improving coverage of its own region, an AI lab commissioning whatever it needs to train on. Each request is placed through the same protocol and fulfilled by the same collector — the federated ISP collective — and every result accumulates in the same dataset. Each funder therefore improves the data for everyone else as a byproduct of serving itself, and the cost of keeping the commons fresh is spread across many parties with unrelated interests rather than resting on any single one.

Once these datasets are open to everyone, holding data stops being an advantage in itself, and the contest moves to what gets built on top of them. The record itself — what the web contains, what the streets look like — also stops being kept by companies whose revenue depends on which version of it people are shown.

In depth · 03

Maintaining the commons is real work — collection has to run, anonymization has to hold, quality has to be checked, storage has to be paid for — and the system treats it as work rather than charity. Every commissioned refresh routes payment to the operators who performed it, and because demand for fresh data never ends, the income arrives not as a windfall but as a recurring line of business: the second revenue stream the ISP acquires in this story.

Part of that stream flows onward to the households whose activity, in aggregate, made the data worth anything. Here the design bows to human nature. People will not run wallets to collect cents; the worth of any one person’s signal is genuinely small, and paying it out directly feels like a chore dressed up as an opportunity. Folded instead into the bill that already exists — as a visible discount, applied automatically, for everyone — the same money becomes something a household actually notices. The operator does the harvesting; the subscriber simply pays less.

In the aggregate, this is the revenue that replaces advertising as the patron of the everyday Internet. Search, mail, news and video have always been paid for by something. Financed from data and computation sold openly, rather than attention sold covertly, the free tier survives with its incentives repaired: the services need their users satisfied, not captivated.

Act III

ISPs becoming app distribution companies

In the ESP model an application is a deployable package. Founders build the product; ISPs and their datacenters run it as local franchises, close to their own subscribers. The protocol splits every cleared payment automatically — operational revenue to the operator, royalty to the founder.

01

New kind of apps break the old trade-off between convenience and control

These apps feel as fast and polished as anything the big platforms offer — yet nothing is taken in exchange. Your data stays under your control and portable: no signup walls, no forced ads, and nothing that ties you to a single provider. And the apps are natively jurisdiction-aware — each user is served by a legally accountable operator in their own country, under their own laws, without forking the product.

Click for in depth

02

Small founder teams win — distribution, billing and compliance come with the network

Freed from the operational overhead, a compact team reaches millions and competes on product quality alone — operators everywhere deploy, market and support its app. Alongside the founders a new class of entity emerges: local Site Operators who run catalogues of apps, specialize in finance, media or AI, and carry the legal weight in each jurisdiction.

Click for in depth

03

Monetizing routing decisions is a new revenue stream for ISPs

Because the ISP controls the routing tables, it decides which site serves each request. It can host the app on-net to capture the full operational revenue, or steer that traffic to partners who pay for the feed. Distribution itself becomes a line of revenue, not just a cost.

Click for in depth

In depth · 01

Decentralized software has historically asked its users for a sacrifice. Systems that pushed every action through global consensus were slower, clumsier and poorer than their centralized rivals by construction — and no principle survives a ten-second wait to post a message. For open architectures to win, they had to stop being a compromise.

The resolution is a division of labor. Only what genuinely requires incorruptibility — money, settlement, ownership — touches the shared ledger. The applications themselves run in real datacenters with everything co-located and tuned, exactly as the incumbents run theirs, so nothing in the experience betrays the architecture beneath. The difference is confined to who holds what: the user’s data lives in storage the user controls, replicated close to the application for speed, readable to it but never absorbed by it. Walking away means pointing that data at a different provider — not reconstructing a digital life from scratch.

Law is treated with the same seriousness as latency. Each user is served from a site operated by an accountable company in their own country, under their own rules; whatever must differ between countries is configured locally rather than forked from the product. The result reverses the terms the platform era imposed: software that feels global and answers locally, convenience without captivity — and nothing holding the user in place except the product continuing to deserve them.

In depth · 02

What has always been scarce for a founding team is not the ability to build software but everything surrounding it: reaching users, charging them, satisfying a hundred regulators, staffing support at three in the morning. This model lifts all of it off the builder’s desk. The creators’ claim on their work is written into the settlement layer itself — a share of every paid interaction, anywhere on earth, flows to them automatically — which means the code can be given away freely without giving away the living it earns.

The operations land with those equipped to carry them. Local operators deploy the application in their own datacenters, answer to their own regulators, staff their own support — and because they earn from every unit of usage they serve, they promote the product to their subscribers as energetically as a franchisee promotes a brand. A handful of engineers thereby inherits, without hiring anyone, the sales force and operations of a multinational.

Even the financing bends to the same logic. A team can raise money against the usage its product will one day generate, repay it automatically from revenue as it arrives, and emerge unencumbered once the agreed cap is reached — no board seats surrendered, no permanent tax on the product’s future. What disciplines the whole arrangement is the absence of walls: builders who neglect their work can be replaced, and builders who keep deserving their share keep receiving it. Usefulness, continuously renewed, is the only tenure.

In depth · 03

Applications in this environment are reached through addresses that many datacenters announce at once, leaving the network to decide which one answers each request. That decision belongs to the ISP, which controls the routes its subscribers’ traffic takes — and here the decision carries money, because whoever serves the request earns the revenue for operating it.

Both available choices pay. An operator that hosts an application locally keeps the full operating income, serves its subscribers at the lowest possible latency, and never pays to carry the traffic elsewhere. An operator that would rather not run a particular service still holds something of value — the demand itself — and can direct it to partners on agreed terms. Between hosting and steering there is no neutral setting in which the position earns nothing; the operator chooses only in which currency to be paid.

Practiced at scale, this becomes a business of portfolio judgment. Capable operators run catalogues of applications on one shared pool of machines, shifting capacity toward whatever is busy tonight and renting the idle remainder out for background work; over time they specialize around their strengths — one excelling at regulated financial services, another at media, a third at machine-learning workloads on its accelerator fleet. Distribution, a function ISPs have performed unpaid since the network began, matures into their third business — and the one that feeds the other two.

Endgame

ISP-centric Internet

Thousands of federated providers — not one platform — hold the Internet’s liquidity, its data commons and its distribution. Together they form an economic force with stronger network effects than any hyperscaler, and metered pay-as-you-go becomes the default business model of the Internet.

01

Hypercentralization ends — applications can no longer build moats, only better products

Every Internet giant was built on the same two moats: captive distribution and hoarded data. In the endgame neither can be meaningfully built, because the federation always holds more of both — no platform can out-distribute a network that owns the last mile to every user, and no private silo can out-collect a commons aggregated across all of them and open to everyone. What remains is competition on merit: product quality, service quality, price. Positions stop being defensible by structure and must be re-earned in use — Metcalfe’s law stops minting monopolies and goes back to work for the network itself.

Click for in depth

02

Economic activity in both directions feeds a neutral wealth fund — a moat no company can cross

The fund is fed from both sides — households’ bills and buyers of data and compute flow in; per-use payouts to builders and discounts back to households flow out. That circulation makes it less a treasury than a monetary system: the place where the Internet’s economy clears. It decouples the price of access from the cost of service, funds the everyday “free” Internet honestly, and gives the long tail of software a stable income no ad market would ever grant it. And it cannot be out-competed, because neutrality is the one thing a company cannot copy — a private replica of the fund is just another walled garden, and walled gardens repel the very flows that make the pool deep. Like a reserve currency, it wins not because anyone defends it, but because everyone’s economics come to be denominated in it. To beat it you would have to out-pool the planet; to out-pool the planet you would have to be neutral — that is, become it.

Click for in depth

03

Jurisdiction-native decentralization — decentralization that faces the law instead of fleeing it

The old Internet was structurally offshore: global platforms served every country from nowhere, and enforcing local law against them meant a decade in court, if it was possible at all. ESP makes jurisdiction a protocol primitive — settlement clears only between parties matched in the same jurisdiction. A German user is served under German law by a locally accountable operator; compliance stops being fuzzy risk management and becomes a binary, deterministic state. Nation-states regain real oversight of their digital economies — a law passed at home is enforced by the network itself, because breaking it severs revenue rather than triggering a lawsuit. Jurisdictional arbitrage stops being a business model, and the digital economy contributes to the society it operates in instead of extracting value to havens.

Click for in depth

In depth · 01

Every dominant platform of the last generation was constructed from the same three materials: control of the route to users, custody of their identities, and exclusive accumulation of the data their activity produced. The products and brands were the visible surface; those three holdings were the load-bearing structure. The end state described here is simply the condition in which none of the three can be held any longer.

The route to users belongs to the federated network and is priced the same for everyone. Identity stays where it naturally lives — with the provider a household already knows — which vouches for its subscribers toward services without handing over who they are; no application accumulates accounts, because none needs them. And data, flowing into open aggregates that anyone may use, can no longer be hoarded into an advantage: whatever one company learns from the commons, its competitor can learn as well. The raw materials of monopoly are still produced, more abundantly than ever; they simply no longer accumulate anywhere private.

A market stripped of these instruments has one axis left, and it is the one users always wanted: quality. A service that decays or turns against its users can be walked away from, data in hand, and every operator has reasons to make the walking easy — though never free, so that building well still compounds. Scale still matters and networks still concentrate; what they now concentrate is the shared fabric itself, which belongs to no one and therefore threatens no one.

In depth · 02

At the center of this economy sits a single shared fund, and everything passes through it in both directions. Households’ payments flow in, alongside the money paid by buyers of data and computation; payments to applications flow out as usage settles, and discounts flow back to households. It is less a treasury than a circulation — the point at which the Internet’s economy clears — and its reserves blend every kind of revenue the network produces.

The blending is what gives it power. Because connectivity fees, service usage, data income and compute income mingle in one pool, no service has to be priced by its own costs alone: the heavy user of one thing is carried by the light user of another, expensive months are absorbed by quiet ones, and the everyday services that anchor the Internet can be offered generously while strict metering runs invisibly below. It is the mechanism of an insurance pool, applied to an entire economy.

What protects the fund is not its size but its constitution. Any imitation built by a company would have an owner, and an owner is exactly what such a pool cannot survive — no operator will pour its subscribers’ economics into a rival’s coffers. Depth comes only from a neutrality that no company can offer, which is why the fund cannot be competed with, only joined. A position held that way needs no defending; it is simply final.

In depth · 03

The legal signature of the platform era was placelessness. A service could be consumed in one country, operated from a second, incorporated in a third and taxed — if at all — in a fourth; every attempt to govern it slid off toward some other jurisdiction. This was not a failure of regulators but a property of the architecture: where the money and the operation can always move, the law can only ever chase.

The new arrangement removes the escape rather than sharpening the pursuit. Settlement is possible only between parties under the same law: whoever serves a user must be a company established where that user lives, and where that condition fails, the payment simply never clears. Obligations that once required treaties to enforce — consumer protection, taxation, lawful process — now attach automatically to someone within reach. And the operators gain as much as they concede: a local company serving local customers under familiar law is a far safer thing to be than a link in an unaccountable global chain.

For governments, this restores authority without expanding surveillance: rules made at home bind the companies at home, and breaking them interrupts revenue directly — a consequence measured in days rather than in decades of litigation. For the Internet as a whole, it ends the era in which regulatory escape was a business model. What is earned in a society settles there, is taxed there, and helps carry the society it came from.

Reinventing the ISP

ISPs now: connectivity companies

Centralization was the winning architecture of the last era

ISPs run the Internet’s one irreplaceable layer, and capture none of its value

The Internet runs on ads or subscriptions; most people fit neither

ISPs becoming clearing companies

Existing connectivity bill is used to settle all Internet service usage

Clearing Internet service usage is a new revenue stream for ISPs

Metered approach to Internet service usage liberates users from subscriptions and advertisements

ISPs becoming data companies

ISPs make up the largest, widest data ingest in the world, paving the way for the largest datasets ever created

Ingested and aggregated data become public commons datasets, not private data silos

Data aggregation work is a new revenue stream for ISPs

ISPs becoming app distribution companies

New kind of apps break the old trade-off between convenience and control

Small founder teams win — distribution, billing and compliance come with the network

Monetizing routing decisions is a new revenue stream for ISPs

ISP-centric Internet

Hypercentralization ends — applications can no longer build moats, only better products

Economic activity in both directions feeds a neutral wealth fund — a moat no company can cross

Jurisdiction-native decentralization — decentralization that faces the law instead of fleeing it

Read Reinventing the ISP