FAANG worked as an acronym for fifteen years because the names on the list controlled the same thing: consumer attention. Five companies, one moat, one revenue model.
That structure is over.
The new acronym circulating across tech this week, MANGOS, swaps half the names and changes what the names actually represent.
Meta. Anthropic. NVIDIA. Google. OpenAI. SpaceX. Some of these companies make models. One makes the chips the models run on. One makes the rockets that carry the bandwidth. Two are not even public.
For the C-Suite reader, this is a map of who your AI strategy is now tethered to, whether you signed up for that exposure or not.
What Actually Changed
FAANG controlled distribution. MANGOS controls production.
Distribution layers compete with each other for the same end user. Production layers compound on each other.
NVIDIA sells to OpenAI.
OpenAI's models run in Google Cloud.
Anthropic trains on the same silicon.
Meta open-sources to weaken the closed providers.
SpaceX builds the satellite backbone that determines who can deploy these models at the edge once terrestrial infrastructure is saturated.
The SpaceX inclusion is the one that catches most readers off guard. It belongs because the next generation of enterprise AI deployment, the one your roadmap is quietly assuming, depends on bandwidth that does not yet exist on terrestrial infrastructure. Satellite backhaul is becoming a strategic input.
What you are looking at is six companies forming an interlocked stack where every layer needs the layer below it.
When your CIO presents the next AI initiative to your board, the cost structure of that initiative is determined by decisions made inside two or three of these six companies. Most boards do not yet see that dependency clearly. Most CIOs are not yet pricing it into their planning horizon.
The Concentration Risk Most Boards Are Not Tracking
When I worked at Microsoft on the Bing Ads revenue engine, the competitive question for the team was a distribution question. How do we win against Google search. The answer involved sales motion, ad product, and pricing.
In the AI era, the equivalent question is a stack question. Which of the six layers are we exposed to, and what happens to our cost base if any one of them moves.
An exercise for the next executive offsite: list every AI tool your organization is paying for, then trace each one back through the MANGOS stack. The number of times your name lands on NVIDIA silicon, OpenAI or Anthropic models, and Google or Microsoft cloud will surprise the room. The concentration is already in your P&L.
The boards I have walked through this exercise with consistently find the same pattern. What looked like vendor diversity at the application layer turns out to be single-vendor exposure at the infrastructure layer. The CFO has been signing off on it without anyone naming it that way.
What the Acronym Is Actually Telling You
Acronyms reflect where power has consolidated. FAANG named the companies that controlled what consumers saw. MANGOS names the companies that control what enterprises can build.
If your competitive strategy depends on AI capability that lives inside this stack, three questions are worth raising at the next board cycle:
Which layers of the stack are we exposed to, and is that exposure single-vendor or distributed?
What happens to our unit economics if pricing changes at any single layer?
Where would we want to build optionality, and at what cost?
The CIO has the inputs. The CFO has the model. The board needs to see both.
Six companies now sit underneath the cost basis of your AI strategy. Most boards have not yet asked which two or three they are most exposed to.
The conversation is already running on LinkedIn. Worth seeing how your peers are reading it: LinkedIn post - MANGOS