MANGOS is only the opening move in AI

A new acronym is gaining traction in boardrooms and investor circles: MANGOS, shorthand for Meta, Anthropic, NVIDIA, Google, OpenAI and SpaceX. It is being positioned as a possible successor to FAANG—a symbol of where technological influence sits in the AI era.

But according to the Founder and Co-CEO of VTEX, MANGOS is not the destination. It is the opening move. Over the next five years, companies that appear structurally unassailable today may lose relevance, decline or be replaced by new leaders, while unexpected challengers emerge from markets few people are watching.

The pace and depth of change will be greater than in previous technology shifts. The companies most likely to win will not simply add AI to existing products and processes. They will rebuild around it.

Why adding AI is not enough

A common approach to AI transformation is to deploy a copilot, add a chatbot, automate a few workflows and report efficiency gains to the board. That may improve productivity, but it does not create an AI-native company.

Recommended reading

Jibo’s successor wants to turn life into AI memories

The more consequential question is: if the company were being created today, with AI as a native capability rather than a retrofit, what would it look like? The answer is usually very different from the organization that exists now. The shift is not just technological; it is a redesign of the company itself.

An AI-native mindset has several structural requirements:

• Rebuild the organization around AI outcomes. Traditional hierarchies were designed for information scarcity, when value came from controlling data, expertise and decision-making authority. As information becomes abundant, judgment, creativity and speed of execution become more important. Reorganizing for AI is not simply about cutting headcount; it is about placing human judgment where it creates the most value.
• Increase talent density deliberately. Adoption will not happen organically at the speed this transition demands. AI training must be treated as a CEO-level mandate, properly resourced and measured against real outcomes—not as an optional employee benefit.
• Eliminate before automating. Organizations often automate inefficient processes without questioning whether those processes should exist. Automating a broken workflow at scale only produces broken results faster. Removing unnecessary approvals, reporting layers, workflows or product features can be more valuable than speeding them up.
• Tie investment to commercial results. AI spending must have a clear path to measurable revenue growth or profitability within a timeframe the business can hold itself accountable for. Technology enthusiasm cannot become its own justification.

Incumbents face an AI-native challenge

In commerce, the distinction is already emerging. The winners will not merely apply AI to search, customer service, merchandising and operations. They will rethink decision-making across product discovery, fulfillment, customer experience, media and profitability.

Market leaders face an uncomfortable obstacle: the processes, organizational structures, institutional knowledge and partner ecosystems that helped them dominate can also slow transformation. Without being questioned, those strengths become anchors.

New entrants have fewer such constraints. They can build AI-native from day one, without legacy infrastructure to protect or internal politics to navigate. They may not announce themselves until they are ready, and by then the gap with incumbents could already be substantial.

The MANGOS companies are building much of the infrastructure for the AI era, but infrastructure providers and application-layer winners are rarely the same businesses. Railroads enabled entire industries without owning all of them, just as internet infrastructure companies in the 1990s did not capture most of the value created online.

MANGOS is building the rails. The larger question is what enterprises will build on top of them—and whether they can redesign fast enough to compete.