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)