Why is crypto mining banned while compute jobs are allowed?

Proof-of-work mining does not run the GPU to produce a defined output from a defined input; it burns energy in proportion to the number of competitors in a hash race. Compute jobs — AI training, rendering, simulation, data processing — produce a measurable, usable output in every GPU hour.

From a business-model perspective the difference is deep: one is a speculative race whose infrastructure cost exceeds the price, the other is productive work where every rented hour reflects real value.

Mining revenue swings with the exchange rate and network difficulty, so there is no guarantee it stays above the hourly rent; a miner in loss abandons the provider’s resources and leaves behind debt and refund disputes. Mining also runs at full power 24/7, putting more stress on cooling, power infrastructure, and rack safety than any other workload on the same GPU.

There is also an abuse dimension: accounts opened with stolen payment instruments, anonymous operators avoiding identity verification, and short-lived mining camps financed with payments that are later charged back are a common abuse pattern. These signals make platforms cautious enough to ban mining outright.

First, a clarification: in Turkey, crypto mining is not a crime by itself; the ban is largely a commercial choice by platforms. But the choice is not arbitrary: actions such as unauthorized electricity use are criminally punishable, earnings are subject to tax law, and high power density exposes the operator to regulatory scrutiny.

On CoreHu GPU, cryptocurrency mining is explicitly prohibited in the terms of service. The reason is an operational boundary, not a legal opinion: keep resources allocated to productive workloads and offer every tenant predictable pricing.

A compute job has three properties: the work is defined (what you run is known), the output is measurable (a model, a render, a simulation result), and it is contract-bound (the acceptable-use scope is written down). The platform takes its cut from the value produced; it is a measurable service exchange, not a gamble.

The same GPU hour becomes a model in AI training, a frame in rendering, or a result set in simulation. That diversity means every tenant filling capacity gets output from the platform without pushing prices up.

• Allowed: AI training and inference, rendering, scientific computing, simulation, data processing.
• Prohibited: proof-of-work mining, jobs that generate attack traffic, proxy/anonymization relays, and any other use that violates the law.

The philosophy of the distinction in one sentence: productive work is free, and resource-heavy speculative work is banned. This policy keeps prices permanently low and predictable for every remaining tenant.