Microsoft shows that Copilot is beginning to monetize
Microsoft reported quarterly revenue of $90 billion and strong growth in its cloud. The most important sign for companies: Microsoft 365 Copilot surpassed 30 million paid users, while Azure continued to grow strongly.
What it means for your company: AI tools for office, documents, emails and reports are already moving from experiment to real work product. But paying licenses is not enough; You have to define specific tasks and measure whether they save time.
Meta accelerates personal and business agents
Meta reinforced its vision of personal AI agents and also agents for enterprises within its ecosystem. Recent reports indicate weekly business usage on WhatsApp, Messenger and Instagram by more than one million businesses.
What it means for international markets: WhatsApp and Instagram will continue to be key channels to sell and serve. The opportunity is not to “put in a bot,” but to design a flow that captures data, qualifies queries, and delivers opportunities to the human team.
Atlassian puts limits on team AI spending
Atlassian began monitoring internal AI usage with monthly budgets per employee. The move comes as several technology companies review whether using AI “for everything” really produces more value or just raises the bill.
Practical lesson: A small business doesn't need to spend uncontrollably to take advantage of AI. It is advisable to define a budget, use cases and rules for when to use expensive models and when to use simpler alternatives.
The trend of spending tokens without measure begins to slow down
AP reported that several companies are looking for cheaper, more disciplined AI after a stage of “tokenmaxxing”: using powerful models for too many tasks, even if the result did not justify the cost.
What it means for an SME: Not everything needs the most powerful model. The correct strategy is to route tasks: the simple with economic models, the complex with advanced models, and always with human review where necessary.
NVIDIA and the Linux Foundation promote a security alliance
NVIDIA, the Linux Foundation and dozens of companies launched the Open Secure AI Alliance to create open tools to help test, audit and govern AI agents and systems.
Why it matters: The conversation is no longer just which model responds best. For companies, identity, permissions, registrations, isolation, auditing and control of agent behavior matter.
The debate on open models and open weights grows
Large technology companies are defending open models as part of security and innovation, while other actors warn of risks of misuse. The debate intensifies because it also affects costs, supplier independence and data control.
Ezelero reading: It is not about choosing a technological religion. It's about choosing the right tool for each process, based on cost, privacy, quality and risk level.
Conclusion for companies across different markets
The underlying news is not that one large company made more money or that another spent more on servers. The useful news is this: enterprise AI is entering a more mature stage, where efficiency, control and actual implementation matter.
For a international SME, the path remains the same: start with a specific process, measure results, control costs and maintain human decision at critical points.
Sources consulted: This page summarizes recent news and explains it in Spanish for the context of companies across different markets. The original links may be in English.
Do you want to apply AI with cost control?
We review your process, choose a specific opportunity and define which tool is best to use without paying more.