The field of AI is shifting from using a large number of in-house dedicated models to using a small number of large multi-task models owned by external companies, and new types of AI projects have emerged - some of which have been developed by external model providers (OpenAI, Anthropic, Meta, DeepSeek...). Teams of non-AI experts can now integrate AI solutions with relative ease, and AI literacy is becoming increasingly important in companies.

Generative Artificial Intelligence (GenAI) opens the door to shortening development cycles, reducing technical and maintenance efforts, and enabling innovative use cases that previously seemed out of reach.

Gartner forecasts that global GenAI spending will reach $644 billion in 2025, up 76.4% from 2024. Hardware such as servers, smartphones, and personal computers (PCS) integrated with AI capabilities will be the main drivers of GenAI spending growth, with 80% of GenAI spending going to hardware.

In 2025, GenAI will see significant growth in spending across all core markets and segments (see Table 1). GenAI will have a huge impact on all aspects of the IT spending market, indicating a future in which AI technology will increasingly become an integral part of business operations and consumer products.

John-David Lovelock, distinguished research vice president at Gartner, said: "The high failure rate of initial proof-of-concept (POC) efforts and dissatisfaction with current GenAI outcomes have led to reduced expectations for GenAI's capabilities. Still, base model providers are investing billions of dollars each year to improve the scale, performance, and reliability of GenAI models. This conflict will continue through 2025 and 2026."

The growth trajectory of the market is heavily influenced by the increasing adoption of AI devices, which are expected to account for almost the entire consumer device market by 2028. However, consumers are not actively seeking out these AI devices. As manufacturers embed AI as a standard capability in consumer devices, consumers will have to pay for it.

At the same time, GenAI also introduces new risks, such as hallucinations and reliance on third-party apis. GenAI's success relies on robust evaluation, as these solutions have general "free output" relationships (answer relevance, correctness, tone, illusion, risk of harm...). Therefore, it needs to be evaluated from multiple perspectives. It is important to run this step before deployment, often referred to as "offline evaluation," but once a GenAI solution is deployed, monitoring becomes critical to ensure it is working the way it is expected and expected in the long term.