Summary of "How to get up to 2 TERABYTES of RAM and / or 224 CPU CORES on a vps for free ?"

Summary

The video claims a method to get temporary, free access to very high-end cloud compute via an Intel Developer Cloud trial environment—framed as an educational AI/compute use case. It repeatedly warns against prohibited activity (especially mining).

It mentions very large resource ceilings (e.g., up to ~2 TB RAM and/or very high CPU core counts—headline figures include up to ~224 CPU cores, though subtitles are inconsistent).

1) Account setup / trial access (high level)

Create a new Gmail account (the video mentions using Cloudflare email forwarding).
Set up/activate an Intel Developer Cloud user account tied to the creator’s workflow.
Use a provided link to request “free access.”
Include details such as:
- Business/institution name
- User type (e.g., media & graphics developer)
Wait for approval to receive “Intel developer cloud access.”
The trial provisioning is described as enabling up to ~3 months of sandbox/session time to run prototypes/experiments on Intel hardware.
Explicit reminder/disclaimer:
- Mining is prohibited and violates terms of service.
- If mining-related detection occurs, the video suggests refreshing/checking registration status.

2) Selecting AI compute resources (core idea)

After login, the guide focuses on AI-related toolkits/samples and uses OpenSSH on Windows:

Copy an SSH key
Configure SSH access:
- Edit/create a user SSH config
- Paste a provided config snippet into the terminal/SSH config file
Run verification commands to check server specs.
The initial allocation is described as small at first (e.g., ~“10 cores/calls” and low RAM initially).

3) Job submission model for provisioning large instances

The key mechanism is job submission using Intel Cloud’s documented commands:

The video indicates multiple server types are available (with different RAM sizes and CPU options).
To request compute:
- Copy a command
- Adjust parameters to target the desired hardware profile
They demonstrate switching to an interactive mode to execute commands inside the provisioned server.

4) Requesting the “maximum” configuration (claimed outcomes)

Important notes/disclaimers in the workflow:

Allocations are not 24/7
The environment is for job management, AI training, and experiments

They demonstrate changing the request target:

From a GPU-focused profile to the intended high RAM / high core type

Reported outcomes (with subtitle inconsistencies):

They initially claim large allocations such as “1,24 GB” / ~“512 GB” (narration/subtitles appear inconsistent).
They then attempt again for:
- “2 terab” / “2 TB RAM”
Final claims include:
- ~2 TB RAM
- ~160 CPU cores
- Earlier headline goals mention up to ~224 CPU cores, but exact figures appear unreliable due to subtitle errors.
CPU hardware is described as Intel Xeon / “Intel … Platinum” (subtitles likely garbled), implying Intel Xeon-class systems.

5) Running code / environment limitations

The video describes several practical limitations and workarounds:

Pipes don’t work as expected (“pipes don’t work the way they expect”).
Python availability is described as very old:
- Python 2.7.8 is mentioned
Workaround:
- Use an installation file provided via their Discord to auto-install:
  - Python
  - noVNC
  - additional packages via a conda environment
- The approach is presented as avoiding typical sudo/pseudo-permission issues, enabling package installs within the VPS.

6) Product/testing cautions & messaging

The video repeatedly includes disclaimer-style messaging:

Use for educational AI/learning
Avoid illegal activities, particularly mining

It ends with typical engagement/CTA messaging:

Like/subscribe call-to-action
References providing the automation/install file through Discord

Main speakers / sources

Speaker/creator: “Fox y” (referred to as “foxy” / “this foxy”)
Service/source used: Intel Developer Cloud (trial sandbox + interactive SSH/job submission)