Engaging Introduction
In 2026, artificial intelligence is no longer an option but an essential budget line for companies. AI spending is skyrocketing, driven by models like Claude from Anthropic, which are already generating an annualized revenue of $14 billion. Yet, paradoxically, these investments are reducing overall operating costs by replacing software and employees. This analysis breaks down the precise figures, sector disruptions, and trends for 2026.
Companies are shifting from a payroll expenditure model to optimized AI budgets, achieving net savings despite the acceleration of investments. The traditional software sector is trembling, while AI cybersecurity is soaring to $308 billion globally. Here’s a breakdown of the real costs.
Explosion of AI Budgets: Key Figures for 2026
Companies are massively reallocating their budgets towards AI. A typical firm that previously spent $100 million annually on salaries and $5 million on AI is now spending $70 million on employees and $20 million on AI. The result: operating costs down by 5%, but the AI budget has quadrupled.
Anthropic exemplifies this trend with an annualized revenue of $14 billion in February 2026, compared to just a handful of premium clients in 2024. The company now has 500 clients spending over $1 million per year. Its valuation has reached $380 billion, boosted by features like Claude’s connection to customer databases for prospecting and automated follow-ups.
The enterprise software sector, valued at $300 billion in annual revenue, sees AI's contribution rise from 2% of revenue in 2025 to a projected 30% by 2035. This substitution is disrupting business models: AI agents are billed on a flat rate, per task, or per request, directly linked to data center computing costs.
SaaS Disruption: Stock Market Plunge and Restructuring
The announcement of new features for Claude on January 30, 2026, wiped out $300 billion (€253 billion) in market valuation for giants like Salesforce, SAP, Oracle, Adobe, ServiceNow, Intuit, Tyler, and Zendesk. The "software and services" values of the S&P 500 have dropped by 27% since October 2025, with a price-to-earnings ratio at its lowest in thirty years.
In France, 40% of traditional software publishers face extinction, as 30% to 80% of their functionalities can be replaced by AI. Non-AI firms could lose 15% to 25% of annual revenue. SaaS acquisitions are exploding: up 26% in the third quarter of 2025, with 746 transactions, a record.
Companies are negotiating hard: traditional SaaS providers, weakened by stock declines of 40% to 60%, can no longer impose annual price increases of 5%. Artefact, a consulting firm, has developed an in-house solution with AI that replaces four paid HR management software (training, resources, vacations).
Comparative Table: SaaS Costs vs AI in 2026
| Provider/Model | Monthly Price per User (USD) | Key Features | Market Impact 2026 | Pricing Evolution |
|---|---|---|---|---|
| Salesforce | $25-300 (per employee) | Traditional CRM | -27% since Oct. 2025 | Towards flat/task |
| Claude (Anthropic) | Flat rate per request (not public) | Prospecting, follow-up | Valuation +$380B | 500 clients >$1M/year |
| SAP | $100-500 (per module) | Legacy ERP | Loss in $300B tsunami | Forced adaptation |
| Zendesk | $49-215 | Customer support | -27% S&P 500 | Model disrupted |
| In-house AI Solution | Marginal cost (electricity) | Complete HR | Savings x4 software | Zero subscription |
This table highlights the shift from fixed subscriptions to variable costs linked to AI computing, more aligned with actual usage.
AI Cybersecurity: $308 Billion in 2026
Cybersecurity, boosted by AI, reaches $308 billion globally in 2026, up 11.8% year-on-year. Security software dominates with over 50% of the market and a growth rate of 14%. The priorities include identity/access management, endpoint security, and security analytics.
Unified AI platforms (CNAPP, cloud-native protection) attract half of software investments. The United States leads with $150 billion, followed by Western Europe ($69 billion) and Asia-Pacific/Japan ($26 billion). The strongest growth is seen in the Middle East, Africa, and Latin America.
Companies are migrating to integrated architectures to counter the complexity of threats, regulation, and the rise of AI. Increased visibility, automated responses, and protection of identities/data in hybrid/cloud environments are key.
-
Advantages of AI solutions in cybersecurity:
-
Automation: reduces response time by 50%.
-
Visibility: +30% on hybrid threats.
-
Cost: unified platforms vs isolated tools, savings of 20-30%.
-
Disadvantages:
-
Data center dependency: rising computing costs.
-
Integration complexity: initial migration costs 10-15% of the budget.
Layoffs and Negative Feedback Loop
AI is improving and becoming cheaper, triggering a spiral: companies lay off workers, reinvest savings in AI, and lay off more. Displaced workers spend less, squeezing margins and accelerating AI adoption. AI budgets increase while overall spending decreases.
For example, Fortune 500 negotiations in 2026 reject 5% SaaS price increases, opting for AI instead. Hyperscalers see their OpEx (operating expenses) investments substituted: no longer salaries but AI computing at the marginal cost of electricity.
In India, TCS, Infosys, and Wipro are losing contracts: AI agents code at the cost of electricity versus low-cost developers. Executive layoffs are rising, and consumption is declining.
Margins and Evolving Business Models
SaaS margins, which have been rising since 2006, risk the end of the "monolithic software." Billing is shifting from subscription/employee to tasks/requests, linked to data centers. Salesforce and Zendesk are already adapting.
Software sector: $300 billion annually, with AI projected to grow from $6 billion (2%) in 2025 to $90 billion projected by 2035 (30%). Small and medium-sized French enterprises: 40% at risk, urgent need for AI acquisitions.
Orange maintains telecom liquidity at €21.3 billion by the end of 2025, with a debt cost of 3.12%. Stability outside pure AI, but exposure through services.
Hidden Costs: Energy and Carbon
AI boosts energy consumption. Carbon prices could peak at €120/t by the end of 2026 in optimistic scenarios. AI data centers: indirect costs through electricity and carbon taxes, up to 10-15% of the AI budget.
Companies are integrating this: AI substitution reduces total OpEx but accelerates dependence on renewable energy.
Outlook 2026-2030: Consolidation and Dominance
2026 marks the peak of disruption: waves of SaaS consolidation, with AI projected to account for 30% of software revenues by 2035. Cybersecurity is expected to reach $430 billion by 2029. Winning companies will be those that migrate quickly to AI agents.
Editorial Opinion Brief AI: The real cost of AI in 2026 is not a budget explosion but a radical optimization. Net savings of 10-20% in OpEx come at the price of a painful transition for legacy SaaS. The priority is to upskill human resources alongside hybrid AI. Watch for: carbon and energy regulation, potential constraints on scaling.