Mobile World Congress in Barcelona is the largest telecommunications-focused conference in the world. More than 100,000 attendees descended on well over 1,000 exhibitors, teasing out all manner of innovative technologies and business strategies.

Although every vendor at the conference struggled to put on its best face, there was nevertheless an undercurrent of desperation, as the much-hyped rollout of 5G has underperformed expectations. And yet, while the telco industry struggles with 5G adoption, there were nevertheless some important areas of innovation worth calling out.

Two areas in particular caught my interest. First, connectivity for the internet of things. Now that the IoT – and edge computing in general – has become a reality, how do enterprises get all their various devices to communicate with each other over vast geographies?

The second area of innovation, unsurprisingly, was artificial intelligence – not AI offerings themselves per se, but rather technologies that support various AI-centric business models.

Here are my pick of the nine most interesting vendors in these two categories – beacons of innovation in an otherwise dim landscape of rising desperation.

IoT connectivity: abstracting the network of devices

The IoT depends upon connectivity, but its challenges are as diverse as its use cases.

Sometimes the devices move around, as with tracking trucks and goods in the logistics industry. In other situations, the devices are stationary and have access to power, but may not have Wi-Fi, as with vehicle charging stations.

Even in factories, where the devices have access to power and Wi-Fi, the sensors may be too numerous and densely packed for Wi-Fi alone to offer adequate connectivity.

Connecting IoT devices with 4G LTE or 5G is often the best alternative to Wi-Fi or wired network connections. In essence, each device has cellphone capabilities that require a SIM (subscriber identity module) card, just as mobile phones do. For geographically dispersed or mobile IoT devices, however, SIM cards have a serious limitation: roaming.

When a mobile device moves from the coverage area of one mobile network operator or MNO coverage area to another, it must take advantage of the roaming agreement between those two operators. However, most MNOs don’t allow devices to roam on their networks for more than three months. Making matters worse, roaming agreements are subject to local regulations – and several jurisdictions forbid the type of roaming that IoT devices require altogether.

Several vendors at MWC addressed these and other IoT connectivity challenges. floLIVE Israel Ltd,, for example, offers a high-performance, low-latency global connectivity backbone that connects to many local MNO networks around the globe.

As a result, it can offer localized connections for every IoT device, enabling them to run on the floLIVE infrastructure with the same SIM card regardless of their physical location.

Eseye Ltd. leverages the floLIVE platform to provide global IoT connectivity via customized device firmware, thus removing the need for physical SIM cards. The Eseye software also provides intelligence within each device, enabling individual devices to choose among available connectivity options while managing and optimizing its connections. Eseye thus works particularly well in situations where different connectivity options may come and go depending on the location of the devices or other factors.

KORE Wireless Group Inc. takes a different approach. Instead of leveraging physical SIM cards or custom device firmware, KORE offers IoT connectivity via eSIM (embedded SIM) cards.

eSIM cards aren’t cards at all – they are pure software that provides virtual SIM capabilities. The primary use for eSIM cards is to give mobile phones a second phone number (I use one in my Dutch phone to keep my old US number active, for example). KORE leverages eSIM technology in IoT devices to provide carrier-agnostic connectivity, with the ability to switch to other protocols as necessary, including satellite-based connectivity.

The next generation of satellite-based connectivity is currently rolling out and will likely be the hot story at an upcoming MWC. Today, however, it’s useful for specific IoT applications like drone connectivity.

Unlike the vendors above, Edgility from Telco Systems focuses on the on-premises edge – IoT networks that have the luxury of gateways that can run in near-edge environments like server rooms, telco points-of-presence, or even data centers.

Edgility provides a Linux-based operating system that runs on any hardware platform, including IoT gateways with Raspberry Pi-level capabilities. In addition, the Edgility software manages and orchestrates each of the Edgility-powered boxes across distributed, global edge deployments far more efficiently than traditional approaches.

These boxes can run business applications as well as virtual routing and firewall capabilities, dramatically reducing the number of boxes that customers require to support complex IoT deployments.

The most innovative of all the IoT connectivity vendors I spoke with, however, is Wirepas Ltd. Wirepas offers IoT device software that can provide peer-to-peer, distributed networking capabilities across large numbers of Wirepas devices.

In other words, Wirepas IoT devices can find and connect to each other across a broad mesh of devices without any other connectivity other than the 5G connectivity that comes with each device.

The company also provides tiny IoT gateways that connect the mesh to the local network or the Internet, giving the entire mesh global connectivity.

A Wirepas mesh offers virtually unlimited scalability, as there is no central point of control handling routing, which means no stateful, memory-consuming routing tables or single points of failure.

Wirepas mesh networks are especially useful for IoT deployments in geographies with spotty cell coverage, for example, smart electric meters in developing countries.

Selling the shovels: vendors supporting AI efforts

You’d have to be living under a rock not to realize that AI will be the hot topic at every conference this year around the globe – and MWC is no exception.

It seems that every vendor had some kind of AI story, with generative AI being the unsurprising favorite topic. Yet for every dozen vendors offering a conversational chatbot experience for whatever widgets they were hawking, there were a few providing differentiated offerings that support their customers’ AI requirements beyond conversational interfaces.

DriveNets Ltd., for example, provides a platform that completely virtualizes network resources. Leveraging commodity “white box” servers, a DriveNets network offers all required network capabilities via a disaggregated, shared infrastructure.

DriveNets’ metal-on-up approach offers more scale and reliability than traditional networks and is thus a popular choice for many different types of telcos and enterprises looking to revamp their data centers.

The DriveNets virtualized approach works especially well with the GPU clusters essential for AI training and inferencing at scale – especially when leveraging large language models.

LLM operations are particularly sensitive to the reliability of the underlying hardware, as a single glitch can require starting over a complex, resource-intensive operation. DriveNets’ high reliability, lossless network offers both solid performance as well as reduced job completion times – essential for LLM-based operations.

Unlike the network-centric vendors in this article, Modulos AG targets AI governance and compliance, especially for regulated industries as well as any European company that must comply with upcoming AI regulations. Though its initial focus is Europe, Modulos can also provide AI compliance for the US-based NIST risk management framework.

The Modulos platform enables businesses to implement responsible AI governance policies while streamlining compliance with changing AI-centric regulations. Modulos gives risk management professionals visibility into AI compliance while also offering mitigation strategies as necessary.

ID R&D Inc. targets another AI-specific problem area: deepfakes. The company’s liveness technology can distinguish between audio and video of real, live people vs. recordings or deepfakes that look and sound like live people but aren’t.

In other words, ID R&D can identify deepfakes – AI-generated human simulacra – as well as differentiating between real and recorded people. The latter situation is the more common problem, as hackers frequently like to fool audio and visual identification applications with recordings or by holding up a photograph to a camera.

Given the preponderance of voice and facial recognition technologies in use today (in particular by government agencies and financial services firms), ID R&D’s liveness detection technology is an important complement to such technologies.

The final vendor on my list is SQream Technologies Ltd. SQream offers acceleration for many different data operations, centering on structured data queries.

SQream’s secret sauce is its ability to run SQL queries on GPUs – a capability that GPUs were never intended for. By accelerating queries on data on its platform, SQream enables its customers to run more queries – and more complex queries – in any given time period.

SQream is thus able to accelerate various data processing and analytics tasks, including several AI-centric ones. Given AI requires the processing of particularly large data sets, SQream can be an essential tool for achieving the goals of AI on time and on budget.

Fear and loathing in Barcelona

There may have been an undercurrent of fear and panic at MWC, as vendors struggled to innovate in a world that seemed reluctant to buy their wares. As this article shows, this simmering desperation was not universal, as some areas are indeed experiencing disruptive innovation that customers actually want to take advantage of.

IoT is one such area – a technology that has taken years to mature but is finally finding its footing. AI as well continues to advance, as the smart money is looking beyond the generative AI hype to the real value beneath.

Jason Bloomberg is managing director of Intellyx BV, which advises business leaders and technology vendors on their digital transformation strategies. He wrote this article for SiliconANGLE. None of the companies mentioned in this article is an Intellyx customer.

Photo: MWC/X