The year 2024 has reshuffled the priorities of the technology sector around three structural axes: local data processing through artificial intelligence, regulatory pressure on the environmental footprint of digital technology, and the rise of quantum cybersecurity. These axes do not operate in silos; they feed into each other and redefine how companies design their infrastructures.
NPU and local inference: the end of all-cloud for artificial intelligence
The most concrete change in 2024 in the field of artificial intelligence does not come from the models themselves, but from where they are executed. Microsoft with its Copilot+ PC, Qualcomm, Intel, Apple, and Samsung are now integrating embedded AI accelerators (NPU) directly into the processors of their devices.
Further reading : The latest online sports news not to miss this week
The principle is simple: instead of sending each request to a remote server, the smartphone or laptop processes image generation, translation, text summarization, or voice assistance locally. This approach addresses two constraints that the cloud alone does not solve: the privacy of personal data and the recurring cost of API calls to hosted models.
To follow tech news on Starlight Infos, this shift towards edge inference also changes how developers design their applications. Reports on software development in 2024 highlight a gradual decoupling: light tasks migrate to the device, heavy tasks remain in the cloud. The result is a hybrid architecture that reduces latency and network dependency.
Recommended read : The latest fashion and culture trends not to miss this season
Digital sobriety and environmental constraints in Europe
Digital transformation is no longer evaluated solely on its technical performance. In 2024, European digital strategies require companies to integrate measurable environmental sustainability goals into their technology choices: energy consumption of data centers, carbon footprint of networks, reuse of existing infrastructures.
This shift directly affects the architecture of IT projects. Three levers are being mobilized by companies anticipating these constraints:
- Optimizing AI models to reduce the computing power required for each inference, thereby decreasing electricity consumption per request.
- Pooling cloud resources among multiple clients or services, instead of provisioning underutilized dedicated capacities.
- Choosing more efficient software architectures that limit unnecessary network exchanges and promote local processing when the task allows.
Consulting firms like Sopra Steria emphasize that environmental sustainability is becoming a criterion for selecting technology providers, on par with price or performance. A company that cannot document the carbon footprint of its digital services loses a concrete competitive advantage during tenders.
The link between NPU and sobriety
Local inference via NPU fits into this logic of sobriety. Each request processed on the device is one that does not pass through a data center. At the scale of millions of devices, the cumulative energy gain is significant, even if each individual operation remains modest.
Quantum cybersecurity: a still theoretical threat but concrete preparations
Quantum computing has not yet produced a machine capable of breaking current encryption algorithms. The threat remains theoretical in the short term. However, preparations to address it are very real in 2024, and the term “cyberquantum” precisely refers to this transitional phase.
The problem can be summarized as follows: data encrypted today with classical protocols could be decrypted retroactively by a future quantum computer powerful enough. This is the so-called “harvest now, decrypt later” scenario. Organizations handling sensitive long-lived data (health, defense, patents) are beginning to migrate to post-quantum encryption algorithms.
This migration does not happen with a single click. It requires auditing all data flows, identifying vulnerable protocols, and then gradually deploying new standards. Companies that delay this audit take a cumulative risk: the greater the volume of exposed data, the higher the cost of a future compromise.
Sustainable technologies and green IT: beyond marketing discourse
Green cloud computing is no longer a communication argument. European regulatory pressure is transforming sustainability into an operational constraint. Cloud providers must now publish precise indicators on the energy consumption of their data centers.
This enforced transparency has a positive collateral effect: it allows IT departments to objectively compare providers on a criterion that was previously opaque. The choice of a host also depends on its documented carbon footprint.
GreenTech and software development
On the development side, the trend translates into the adoption of “software sustainability” practices:
- Systematic measurement of the energy consumption of applications in production, not just during testing.
- Reducing the weight of web pages and mobile applications to limit network transfers.
- Extending the lifespan of devices through lighter software updates compatible with older hardware.
These practices align with the “long-lasting tech” trend identified by the VML Intelligence report The Future 100: designing digital products and services to last, rather than forcing constant hardware renewal.
The technological trends of 2024 share a common thread: computing power is moving closer to the user while environmental and security constraints frame this decentralization. Companies that structure their investments around this triptych (local inference, measurable sobriety, post-quantum preparation) are laying the technical foundations for the next three to five years.