The tech space is often overly concerned with the new and flashy — it seems like every day, there is a new product release, a new iteration, some big new thing that secures headlines and coverage.

But the reality is that there is a whole world of old protocols that are not only alive — they are powering a simply wild amount of the modern internet. The backbone of global trade, banking, energy, and logistics does not run on OpenAI’s latest and greatest AI API — it runs on decades-old infrastructure and protocols that have quietly been revised and improved at scale.

And in an interesting turn of events, many of these technologies are not dying or barely holding on. Instead, they are quietly thriving in the age of AI, finding a new purpose and a new era.

Below, we’ll dive into some of the protocols and systems still quietly running the world.

SOAP Wasn’t Killed by REST — It Just Went Fully Enterprise

Too often, the conversation around REST and SOAP feels a bit too like sports banter. A lot of discourse simplifies things into “REST killed SOAP” or “SOAP is outdated.” While SOAP may be invoked as a punchline in tech circles, in reality, it still powers a shocking number of financial and telecom systems.

Simple Object Access Protocol (SOAP) is a different beast from REST, and all the facets that are typically rendered as negative are actually benefits in certain use cases. The contractual nature of SOAP, and its highly deterministic behavior, is actually critical for stacks that need predictable and compliant systems. When every transaction must be highly controlled and cleared, something like SOAP is a must — and for this reason, it is core to operations like ACH payments, SWIFT message exchanges, loan processing, or financial data connections.

The reality is that REST is not an alternative to SOAP because it is better — it is just very different, and the sporting mentality is shortsighted. While it may indeed be an old tech, it is still trusted and widely adopted.

Example Implementations

You can still find SOAP in any application that needs highly deterministic contract negotiation, including:

• Banking and payment processing: Many core technologies, like ACH, SWIFT, and SEPA message exchanges, rely on SOAP web services for compliance and tracing.
• Telecom provisioning: Many companies use SOAP to manage contractual processing, such as SIM activation, roaming management, and customer billing.
• Enterprise resource planning (ERP): ERPs like SAP and Oracle still provide native SOAP implementations, often as defaults, for their core platforms.

EDI Still Moves Most of the World’s Freight

EDI, or Electronic Data Interchange, was born in the 1970s as a standardized way to exchange business documentation and data between production and logistics partners. In 2025, it operates the de facto backbone of the global supply chain, with orgs like FedEx, Walmart, and massive manufacturing conglomerates depending on EDI to process orders, shipments, and invoices. EDI standards like EDIFACT and ANSI X12 still control much of the API interactions in this space.

And this is not the “last vestiges” of the tech — EDI implementation and management is steadily growing. With AI and automation seeing wider adoption in the industrial space, modern EDI gateways now pipe data into LLMs and analytics systems to do everything from predicting delays to optimizing shipment loading.

In many ways, the structural reality and implementation of EDI make it the perfect structured data to ingest for supply-chain AI — it’s well standardized, widely implemented, and structured for clarity of ownership, processing staging between discrete units or sites, and access controls. This is a goldmine for novel AI logistics implementation, as it cuts out an entire data processing stage that must be done for orgs in other industries.

Example Implementations

EDI regularly pops up in the context of logistics, powering entire industries like:

• Retail logistics: Orgs like Costco and Walmart use EDI widely, especially in terms of shipping notices, purchase orders, and supplier invoicing that is dependent on EDI X12 (850/856/810) transactions.
• Global manufacturing: FedEx, Maersk, and Boeing are high-profile examples of EDIFACT EDI use as standard methodology for inventory and logistics management.
• Healthcare and insurance processing: Notably, the U.S. mandates EDI 837/835-compliant messages for medical claims and remittance transactions.

RPC: What’s Old Is New

RPC was designed to invoke remote functions as if they were local. In essence, you could interact with a mainframe as if you were sitting right at the console, which was an incredibly novel idea at the time. In 2025, you might think that this has been abstracted away entirely — but RPC is actually seeing quite a renaissance, thanks in large part to Google.

While XML-RPC and JSON-RPC have seen reductions in use, modern frameworks like gRPC have seen RPC actually grow in market share, powering everything from Kubernetes microservice clusters to machine learning pipelines. The thing is that RPC allows for a relatively direct connection to local functions and systems in a way that modern AI and ML require. It has rapidly become a de facto way of interacting with systems that require high system access.

The core idea has not changed — RPC is still designed for serialized data efficiency and direct function access. What has evolved is the transport and encoding methodology, with solutions like Protobufs revolutionizing how the tech works.

Example Implementations

• Microservices: Solutions like Kubernetes, Envoy, and Istio use gRPC to manage cluster communication due to its low latency and type-safe schema.
• Machine learning and LLMs: AI requires direct access to serialized model data and inference requests as well as to underlying hardware systems, so RPC offers a great methodology for managing this process.
• Interconnection and hardware APIs: Specific implementations like CUDA RPC allow for direct communication across remote worker clusters without significant data or memory overhead.

MQTT and AMQP Power the Edge

In the past, Message Queuing Telemetry Transport (MQTT) and Advanced Message Queuing Protocol (AMQP) were the kings of the publisher-subscriber paradigm.

Because MQTT and AMQP were really good at dealing with unreliable networks and ensuring completion of message transfer, they both seemed to find their home in each evolution of networking and the internet. In the early days, before broadband was standard, they were the perfect solution for unstable connections.

Then, during the rapid rush to adopt the internet of things, they were the perfect solution for network devices that were not powerful enough to handle their own connections. And currently, they are deploying AI at the edge, streaming live sensor data into on-prem LLMs and managing everything from predictive maintenance to contextual reasoning in production lines.

Ultimately, MQTT and AMQP would be seen as API protocols that were purpose-built for a very specific problem, and that problem — unreliable network connections paired with high data throughput and the need to ensure messages are actually handled properly — did not go away.

Example Implementations

• Industrial IoT: Providers like Siemens and Honeywell use MQTT brokers to transmit telemetry to analytics dashboards and on-prem models for OEE optimization.
• Smart vehicles: Brands like Tesla and Volkswagen use MQTT to stream real-time vehicle diagnostics to handle intermittent connectivity while ensuring optimal operation.
• Enterprise messaging: AMQP underpins solutions like RabbitMQ and Azure Service Bus, serving as the core routing tech for transactional messages in complex microservice environments.

CORBA Is in Everything from NASA to Air Traffic Control

Common Object Request Broker Architecture (CORBA) was first standardized in 1991 as a solution to manage distributed computing across languages, networks, hardware, and deployments. It was essentially an answer to the problems that would eventually inspire microservices, and it caught on quite quickly in industries racing to standardize access to disparate machines.

The use of an interface definition language (IDL) allowed for a sort of standardized communication that was desperately needed in the evolving digital revolution of the 90s, where there was a standard for every bespoke use case, and the machines integrating with one another were often from different eras and built for different purposes. For this reason, CORBA saw wide adoption in solutions like air traffic control, telecom signaling, NASA spacecraft telemetry, satellite tracking systems, and much more.

While modern equivalents like gRPC and complex service meshes are starting to edge into the install base, CORBA is still used quite widely. CORBA was made to be interoperable — and for that reason, it is hard to argue that it is obsolete when it still does its job well.

Example Implementations

• Aerospace: In the U.S., NASA used CORBA significantly in the Space Shuttle and still utilizes it in the ISS ground control systems for deterministic data exchange.
• Aviation systems: Everything from EURONCONTROL to U.S. Air Traffic Management uses CORBA for coordinating data exchange between radar and control centers.
• Telecom infrastructure: CORBA still powers a lot of the legacy 3G and 4G systems and OSS/BSS platforms managing configuration and network orchestration.

What’s Old Is New Again

These are great examples of situations where the age of the tech is not a problem. For many, “if it works, don’t break it” is a solid strategy when faced with migrating a CORBA system to something more modern.

For today’s greenfield development, RESTful APIs that operate over HTTP and use the OpenAPI specification are largely the norm. REST was used by 86% of API developers, according to Postman’s 2023 State of the API Report. This was followed by webhooks, GraphQL, SOAP, and websockets.

While there are some real benefits of moving from old API standards to more modern tech, many industries rely on these systems staying exactly how they are. So, you are likely to see these implementations still in place 10 or 15 years from now.