Which technologies have truly progressed in 2024, and which remain at the stage of marketing promise? Between the massive adoption of generative AI in businesses and the emergence of neuromorphic chips consuming less than one milliwatt, the technological innovations of 2024 paint a contrasting landscape. This article measures the gaps between actual adoption, technical maturity, and environmental impact.
Neuromorphic Chips and Edge AI: The Silent Breakthrough of 2024
Large generative AI models capture media attention, but another family of processors is making progress quietly. Neuromorphic chips, designed to mimic the functioning of biological neurons, allow for the execution of artificial intelligence tasks directly on sensors or embedded devices, without a connection to the cloud.
The Akida Pico chip from BrainChip illustrates this trend: it operates with less than one milliwatt of consumption, a threshold that makes embedded AI possible on battery-powered or energy-harvesting devices. BrainChip has also signed a strategic agreement with Parsons to accelerate the deployment of edge AI systems in the defense sector.
This type of technology is largely absent from most “2024 trends” reports, even though it addresses a major constraint: processing data locally, without network latency and without the energy costs associated with data centers. For companies following the latest innovations on Net Addict’s tech page, this segment deserves special attention.
Generative AI in Business: Massive Adoption, Uses Still Concentrated
The global survey conducted by McKinsey in spring 2024 presents a clear finding: 65% of organizations report regularly using generative AI in at least one function. This figure has nearly doubled compared to the survey conducted ten months earlier.
The details of usage nuance this rapid adoption. Marketing and sales come out on top, with 34% of respondents citing this function as the primary area of deployment. Core business functions (R&D, production, engineering) remain significantly behind.
| Indicator | Result (McKinsey, Spring 2024) |
|---|---|
| Organizations regularly using generative AI | 65% |
| First cited usage function | Marketing and sales (34%) |
| Change compared to 10 months earlier | Nearly doubling of adoption |
The gap between broad diffusion and deployment in high-value functions indicates that most companies are still in a phase of experimentation. Generative AI produces text, images, and code, but its integration into industrial processes or research requires longer adaptations.
Digital Carbon Footprint: Data Contradicting Discourse
The acceleration of AI and the cloud has a measurable environmental cost. A report from the ITU published on 2024 data estimates the operational emissions of the 200 largest digital companies at 301 million tons of CO₂e, accounting for about 1.7% of global electricity consumption.
In contrast, the climate commitments of these same companies have not kept pace with the growth of their infrastructures. The training and inference of large AI models require significantly increased computing power, which drives up the energy demand of data centers.
- The operational emissions of the 200 largest digital companies reach 301 million tons of CO₂e according to the ITU.
- Training a large language model consumes energy amounts that increase with each model generation.
- The “green IT” promises made by industry players have not yet translated into a measurable reduction in overall emissions.
European Technological Sovereignty: A Structural Delay in AI Infrastructure
Europe has research capabilities but suffers from a deficit in the computing infrastructure necessary for AI. An analysis by Bruegel highlights a lack of AI computing capacity on the continent, which hinders the transition from research to production.
This deficit has direct consequences on the market. European companies are heavily dependent on American cloud providers to train and deploy their models. This dependence creates a risk to data sovereignty and limits local innovation capacity in the digital sector.
Conversely, some specialized initiatives are making progress. The Dutch startup Innatera is developing neuromorphic chips designed to operate with a consumption of less than one milliwatt, positioning Europe in the segment of ultra-low-power embedded AI. This niche, less capital-intensive than large language models, aligns better with the structure of the European industrial fabric.
Technologies 2024: What the Gaps Between Promise and Deployment Reveal
The table below summarizes the actual maturity of the main technological trends of 2024:
| Technology | Adoption | Industrial Maturity |
|---|---|---|
| Generative AI | Widespread (65% of organizations) | Concentrated in marketing |
| Neuromorphic Chips | Niche (defense, sensors) | Advanced prototypes, first commercial agreements |
| European AI Infrastructure | Delayed | Persistent US cloud dependence |
| Green IT / Digital Sobriety | Widespread discourse | Emissions on the rise (301 Mt CO₂e) |
The technological trends of 2024 are better understood through their contradictions than through their announcements. Generative AI is spreading rapidly but remains confined to peripheral uses in the majority of companies. Neuromorphic chips open a promising technical pathway, supported by players like BrainChip and Innatera, without yet benefiting from a mass market. The carbon footprint of the digital sector, meanwhile, continues to increase despite the stated commitments.
