The Corporate Case For Decentralization

The corporate cloud contract is the most unexamined infrastructure decision in modern business. A company commits to ten years of dependency on a single provider. The signature takes thirty seconds. The decision behind it took none.

This is the default. Sign with the hyperscaler. Move on. The choice is so reflexive that calling it a choice feels generous. It is what every comparable company is doing. It is the safe path. It is the professional path. It is the path that does not require defending.

It is being quietly abandoned by companies that are paying attention.

Decentralized infrastructure used to be a crypto buzzword. That was a fair read in 2021. It is no longer fair. The largest mobile carriers in North America are running production traffic through community-built wireless networks. The largest aerospace company in the world is storing data on a decentralized file system and deploying nodes of that file system into orbit. National archives, museums, and universities are committing permanent records to decentralized storage as a hedge against single-provider failure. This is procurement that has already happened.

The corporate case for decentralization is not being made by ideologues. It is being made by purchasing departments. It is being made by chief technology officers who have watched their cloud bills double every three years. It is being made by general counsels who read the indemnity clauses and noticed what is not covered. It is being made by operations teams who lost fifteen hours of revenue in October when a single Amazon region went dark and took half the internet with it. The case is not philosophical. The case is economic.

What The Procurement Desk Already Signed

Look at what AT&T did. In April of 2025, AT&T announced a commercial agreement with Helium, a community-built wireless network. AT&T is the second largest mobile carrier in the United States. Helium is a network of hotspots owned and operated by independent individuals and small businesses. The two have nothing in common in terms of structure. They are now in business together.

The structure of the deal matters. AT&T did not buy Helium. AT&T did not invest in Helium. AT&T did not partner with a centralized company that wraps Helium. AT&T integrated directly with a decentralized wireless network so AT&T subscribers automatically connect to community-built hotspots when in range. The hotspots are owned by tens of thousands of independent operators. None of those operators work for AT&T. None of them sign procurement agreements with AT&T. The network functions because the protocol coordinates them, not because a vendor manages them. Mario Di Dio (Helium GM) called it a "commercial agreement," not a pilot. Thousands of locations live across the United States.

Telefónica did the same thing in Mexico. The carrier integrated Helium service into Movistar Mexico, reaching over 2.3 million subscribers, per Messari's State of Helium reports. The full rollout began in 2025. Telefónica is selling Helium hotspots in retail stores. A multinational telecom company is now actively recruiting its own customers into a decentralized network that competes with traditional infrastructure deployment.

The numbers are not small. By the end of Q4 2025, Helium had offloaded 9,840 terabytes of cumulative carrier data, with 4,388 TB moved in Q4 alone (60.7 percent quarter-over-quarter growth). Peak daily active users hit 2.5 million on December 20, 2025. Annualized revenue ran $11.0 million in Q4 (excluding subscriber burn) on top of $18.3 million in Q3. Carrier partners include T-Mobile, Movistar, AT&T, Google Orion, and Wefi. Brazilian Mambo WiFi joined in Q4 with 40,000+ access points.

The reasoning is simple. Building cellular infrastructure is expensive. A 5G rollout in the United States was projected to cost hundreds of billions in private investment. The capital expenditure is the bottleneck. The community-deployed model bypasses that bottleneck. The carriers do not have to deploy the hardware. The community deploys it. The carriers pay for the traffic that flows through it. The capex moves from the corporate balance sheet to thousands of small operators happy to put a hotspot on their roof in exchange for tokens.

Now look at Filecoin. The Internet Archive, the Smithsonian, MIT Open Learning, Lockheed Martin Space, and the Flickr Foundation are storing data on the decentralized storage network. Lockheed Martin has gone further, deploying nodes of the underlying file system into space, using IPFS to coordinate data transfer between Earth and orbital infrastructure where centralized cloud architectures are actively unworkable. Over 800 enterprise clients store more than 1,000 tebibytes of data each on the network. These are production datasets that organizations have decided are too important to leave with any single cloud provider.

What The Centralized Pitch Was Sold On

The centralized cloud was sold on three premises. One throat to choke. Predictable contracts. Operational simplicity.

The first premise is the procurement story. Enterprise buyers were told that the value of buying from a hyperscaler was vendor accountability. If something breaks, you have a phone number to call. You have a service level agreement. You have a contract with a specific company that has a specific obligation. A decentralized network has no CEO. There is no throat to choke.

This argument is intact, technically. There is no CEO of the Helium network. There is no person to fire when something goes wrong. The argument fails for a different reason. The throat-to-choke model assumes the throat matters when something breaks. Hyperscaler service level agreements allow over fifty minutes of downtime per year and cap the penalty at a fraction of the monthly bill. When AWS US-EAST-1 went dark for fifteen hours in October 2025, the contractual remedy was a service credit. The damages to affected businesses were estimated at hundreds of millions of dollars. The credits did not approach that number.

The throat-to-choke model means the throat shows up to a meeting after the outage. It does not mean the throat covers the actual cost. The accountability enterprise procurement was paying for turned out to be expensive customer service.

The second premise is the predictable-contract story. Hyperscaler contracts are well-understood. Procurement teams have decades of experience signing them. The legal language is standardized. The pricing is published. There is comfort in the familiar.

The contracts are not static. Hyperscaler pricing has moved in one direction for a decade, and that direction is up. Enterprise cloud spend has compounded at double-digit annual rates since 2015. Egress fees, the charges for moving data out of a hyperscaler region, function as switching costs that lock enterprises into the vendor they started with. Predictability is a story enterprises tell themselves. The reality is the cost trajectory is predictable in the same way a leak is predictable. You can predict it will get worse.

The third premise is operational simplicity. Hyperscalers run the infrastructure so enterprises do not have to. This was true and remains true. It is also no longer the only option.

The honest case requires naming what decentralization actually asks of an enterprise. It asks for new operational expertise. Enterprise IT teams know how to manage AWS. They do not know how to manage participation in a decentralized protocol. The skill gap is real. The first wave of adoption is happening at companies that have either built the expertise internally or partnered with intermediaries who provide it as a service.

It asks for different procurement language. The contracts do not look like hyperscaler contracts. Accountability sits in protocol design rather than vendor obligation. Legal teams have to learn how to evaluate a protocol the way they used to evaluate a vendor. Some are doing this. Most are not yet.

It does not work for every workload. Real-time financial settlement, ultra-low-latency consumer applications, and certain regulatory-bound workloads are not yet good fits. Decentralized storage works extremely well for archival data, scientific datasets, and long-tail content. It works less well for hot transactional state. Decentralized wireless works for offload and supplementary coverage. It does not yet replace a national 5G build.

The migration costs are real. The savings show up over years, not quarters. The companies adopting decentralized infrastructure now are not doing it because it is the cheapest option in the first month. They are doing it because the cost trajectory of the centralized alternative has become unsustainable and because the operational risks of single-provider concentration have become legible to the people signing the checks.

The Decision Stops Being Automatic

The signature still takes thirty seconds. The procurement playbook still has the same first page. The hyperscaler contracts still get drafted by the same lawyers and signed by the same executives. None of that has changed. What changes is the cost of not asking what else is possible.

The decision is no longer thirty seconds of pen movement. The decision is the most important infrastructure choice the company will make this decade, and it is finally being made on purpose.

The question is not whether decentralization works. The question stopped being that several billion dollars of enterprise spend ago. The question is what the unexamined default costs, and whether the company signing the contract has examined it. The corporate case for decentralization is not a future case. It is a present case. The carriers have already made it. The aerospace primes have already made it. The national archives have already made it. The remaining question is who notices in time to participate in the next decade of infrastructure rather than to pay for the last decade of it.

The signature still takes thirty seconds. The decision now takes longer.