Compute, concentrated.
dense² builds and operates reserved GPU capacity in the Nearshore Americas for US teams. Based in San Francisco, deployed next to the US border.
Why we exist
AI demand is outrunning the power, space and interconnection queues in US metros. Teams that need dedicated GPUs wait for quota, or settle for best effort hosts.
The border metros of the Americas have power, space and fiber one hop from US interconnection hubs. dense² turns that into reserved clusters a US team can contract the way it already buys: fixed topology, a delivery date, named support.
We start in Mexico and replicate the same block across the region.
How we work
Numbers before adjectives
Topology, latency and terms in writing before anything else.
Residency stated plainly
Our capacity sits in Mexico, and every document says so.
Reserved, not best effort
Dedicated hardware on contract, never oversold.
Built against signed terms
We build capacity against contracts, so what we sell exists on the date we say.
Compliance first
Every customer is screened against US export control and sanctions rules.
One standard, replicated
The same 256 GPU block at every site, so nothing gets relearned.
Where we are going
Each border metro pairs with the US market across the line. Only Monterrey is open for contracts today, with 1,024 GPUs planned in the first 12 months.
- Monterrey→TexasLive Q2 2027
- Tijuana→CaliforniaPipeline
- Cancún→FloridaPipeline
- Querétaro→Central MexicoPipeline
- Central America and the Southern Cone→RegionalHorizon
The company
Based in
San Francisco, with operations in Mexico.
Partners
We work with an interconnect partner and colocation operators across the region. They are named in contracts, not on this site.
Contact
[email protected] for capacity, partnerships and press.