Analysis ·
The AI bill soars, yet companies are still looking for a way to measure ROI
Original title, in French: “La facture IA explose mais les entreprises cherchent encore le mode d’emploi du ROI”
- Publication
- IT for Business (France)
- Author
- Laurent Delattre
- Published on
- 1 October 2026 · read in the page structured data
Summary
According to Gartner, global AI spending is set to grow by 49.5% in 2026, reaching $2,670 billion, most of it devoted to infrastructure. A Tokenomics Foundation study of 472 respondents from large companies shows that 73% of organisations struggle to link their AI spending to a business outcome that is measurable for their finance leadership, and only 9% say they are very capable of doing so. The Najar study, also cited, notes that AI is becoming a factor in software contract renewals, with price increases of 8 to 25%, while activation rates rarely exceed 10%. In light of this, a new discipline called "Tokenomics" is emerging to measure and attribute AI costs by workload, much like FinOps applied to the cloud.
Summary based on the full text of the article.
Read the article on IT for Business →Figures cited
-
2,670 USD billion
global AI spending in 2026
global, 2026
« Les dépenses mondiales consacrées à l’intelligence artificielle devraient progresser de 49,5 % en 2026, après avoir déjà atteint 1 787 milliards de dollars l’an dernier. »
Source: Gartner
-
3,637 USD billion
global AI spending in 2027
global, 2027
« Et l’addition pourrait grimper à 3 637 milliards dès 2027. »
Source: Gartner
-
1,484 USD billion
AI infrastructure spending in 2026
global, 2026
« L’infrastructure IA absorbera à elle seule 1 484 milliards de dollars cette année, soit plus de la moitié de la dépense totale. »
Source: Gartner
-
462 USD billion
AI software spending in 2026
global, 2026
« Gartner prévoit 462 milliards de dollars de dépenses en logiciels IA en 2026, contre 288 milliards l’année précédente. »
Source: Gartner
-
73 %
organisations unable to link AI spending to a measurable business outcome
472 respondents, large companies, 11 industries
« elle montre que 73 % des organisations se disent peu ou pas capables de relier leurs dépenses IA à un résultat métier mesurable que leur directeur financier jugerait recevable »
Source: Tokenomics Foundation, State of Tokenomics
-
9 %
organisations very capable of demonstrating AI ROI
472 respondents, large companies, 11 industries
« Seules 9 % s’en déclarent très capables. »
Source: Tokenomics Foundation, State of Tokenomics
-
43 %
respondents citing demonstrating value and ROI as a challenge
472 respondents
« 43 % citent la démonstration de la valeur et du ROI parmi leurs difficultés, de loin la plus fréquente »
Source: Tokenomics Foundation, State of Tokenomics
-
86 %
respondents evaluating or using a model router
472 respondents
« Selon l’étude, 86 % des répondants évaluent ou utilisent déjà un tel dispositif »
Source: Tokenomics Foundation, State of Tokenomics
-
— %
annual AI-related price increase on software contracts
Europe, Salesforce, HubSpot and Zendesk contracts
« avec des hausses annuelles de 8 à 25 % que le cabinet lui attribue »
Source: Najar
-
93 %
use of specialised AI tools outside the contract
« 93 % de l’usage des outils d’IA spécialisés échappe à tout contrat »
Classification
- Country
- Worldwide scope
- Company size
- Large companies
- Industries
- All industries, or industry not specified · Software vendors, IT services companies, telecoms
- Companies
- Gartner (study author), Najar (study author), Salesforce (cited), HubSpot (cited), Zendesk (cited), Tokenomics Foundation (study author)
- Keywords
- AI ROI, tokenomics, FinOps, AI spending, Gartner, AI infrastructure
Entry written by Marsanne Studio from the cited article, with the help of an AI model (claude-sonnet-5). The rights to the article and its illustrations belong to IT for Business and its author. Entry translated from French by Claude Sonnet 5.5.