5G, AI, and dedicated chipsets: the factors powering IoT's second wave

Remember when the Internet of Things was the ‘new thing’? Back in 2010, there was optimism about a world of networked devices. Ericsson predicted there would be 50 billion connections by 2020. It didn’t quite happen. Instead, the IoT followed a more typical technology story. It progressed in some areas, and disappointed in others. The smart home struggled to match the hype. But the industrial IoT delivered real value in manufacturing, logistics, agriculture and more.

So, where is the market now?

According to a new report by IoT Analytics, we are now poised for a decade of growth. The firm attributes this to the impact of AI, along with advances in chip design and connectivity. Another important driver is standardisation. The IoT has moved from fragmented chaos to interoperability – especially in the areas of connectivity, application layer and security. 

Let’s explore the evolution of IoT in more depth.

Boom time – from 21.1 billion IoT devices in 2025 to 55 billion by 2035

IoT Analytics estimates the number of connected IoT devices to have grown 14 percent year-over-year to 21.1 billion in 2025 – around 300 million below previous projections, thanks to capex deferrals and soft demand in China. However, the analyst is confident the total will hit 39 billion by 2030 – and 55 billion by 2035. 

Which network tech is powering the next phase of IoT?

The IoT relies on a variety of technologies to connect devices and sensors. Three account for 76 per cent of all connections: Wi-Fi (32 percent), Bluetooth (24 percent) and cellular (22 percent). 

All three are undergoing changes (in terms of range, battery life, bandwidth and more) that will shape their influence on the space over the coming decade. They have also benefited from cross-industry standardisation.

• Wi-Fi 

Wi-Fi is the looks set to remain the key tech for IoT connectivity, especially for markets such as smart appliances and building automation. Many enterprises are now switching to the Wi-Fi 7 standard (introduced in 2024) for high-speed low-latency applications (Wi-Fi 7 supports speeds up to 46 Gbps). Wi-Fi HaLow (802.11ah) is also gaining traction. It provides long-range, low-power use cases.

• Bluetooth

Bluetooth is the IoT market’s second-largest connectivity type. It is particularly suitable for short-range, low-power applications. The Bluetooth Low Energy ‘flavour’ of the tech continues to lead in battery-powered IoT and in devices that use SoC chipsets to reduce cost and power consumption. Meanwhile, Bluetooth 5.4 is becoming the preferred platform for electronic shelf labels.

• Cellular IoT

After a slow start, the cellular IoT market is gaining market share. Connections grew 16 percent in 2024, and should grow by 23 percent CAGR to 2030. 

For many enterprises, fast and ubiquitous 5G is an attractive alternative to Wi-Fi. It is becoming the standard for high-reliability low-latency IoT use cases such as fixed wireless access, video telematics, industrial gateways, and automotive. It is also key to the construction of private cellular networks, which are now deployed by more than 1,900 organisations across 84 countries.

Meanwhile, with the sunsetting of 2G and 3G networks, LTE-M (Cat-M1), LTE-Cat-1 BIS, and 5G RedCap (Reduced Capability) have emerged as the leading contenders for mid-tier IoT devices. 

Perhaps the biggest driver of Cellular IoT will be the emergence of the SGP.32 standard. SGP.32 is a new protocol for connecting eSIMs embedded into remote IoT device sensors. It is replacing legacy standards such as SGP.02, which weren’t designed for IoT at scale. SGP.32 radically simplifies over the air device provisioning. It means that SGP.32-certified suppliers such as Thales can offer ‘build-once, ship-anywhere’ services to their IoT customers.

The technologies powering a smarter IoT: AI and dedicated chipsets

AI algorithms do more than analyse data. They identify patterns and make decisions/predictions. This has big implications in the IoT – enabling industrial IoT devices to operate with minimal human intervention. But AI requires computational resources, which are typically not available locally. Specialised AI chips are changing this. They allow real-time data analysis and decision-making on the IoT device without relying on cloud connectivity. 

Chip specialists have already launched products that support ultra-low-power AI processing. They include ARM’s Cortex-M55 and Ethos-U55 designs, Qualcomm’s QCS series and NVIDIA’s Jetson line.

Standardisation should help to boost AI chipsets too. Consider the RISC-V open source architecture, which gives developers the ability to design chips without paying royalties. According to RISC-V International, 27 percent of AI accelerators for automotive could be based on RISC-V by 2030.

A unified focus on IoT security

The IoT presents a huge security challenge. It offers hackers a potential attack surface comprising billions of devices, most of which have no human custodian. In the early days, security was an overlooked factor in the design of many IoT devices – and the data they were transmitting. This is no longer the case. Stakeholders are now collaborating to protect the three key factors of: device, network, data. 

A good cross-industry example is the GSMA’s eSIM Security Assurance (eSA) scheme. It ensures that eSIM component (hardware, firmware, OS, and cryptographic libraries) comply with strict security and functional requirements. In so doing, eSA offers a globally recognised 'seal of trust’ for IoT manufacturers.

Regulatory developments: a new era of compliance for the IoT

Another consequence of growing IoT security risks is increased regulation. Governments are not content to let the industry police itself. Instead regulators are imposing safeguards around data protection, privacy, and operational resilience. A key example is the EU Cyber Resilience Act (expected in 2027). It establishes mandatory cybersecurity standards for connected products. For example, IoT devices must ship with secure default configurations, and no default passwords. Environmental regulations will also affect the IoT space. For example, the Ecodesign for Sustainable Products Regulation imposes requirements on the energy efficiency, durability, and recyclability of devices. 

The next phase of the IoT? Intelligence, longevity, reliability

While there could be 55 billion connected devices by 2035, the next phase of IoT growth will not be defined by volume alone. Instead, its key characteristics will be intelligence, longevity, and reliability. Advances in AI, chipsets and connectivity will combine with regulatory maturity to supercharge the space. We will see:

• Intelligence moving to the device and the edge
• Security designed in from the outset, not retrofitted
• Regulation and compliance becoming enablers of trust and scale, not just constraint

Ultimately, the future of IoT is not simply about connecting more devices, but about building systems that are smart, secure, and capable of being trusted over time—delivering lasting value for businesses, communities, and society.