> For the complete documentation index, see [llms.txt](https://docs.aethir.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.aethir.com/aethir-cloud/aethir-cloud-host/operational-requirements-for-cloud-hosts/staking-as-cloud-host/faq-k-value-adjustment.md).

# FAQ: K-Value Adjustment

<details>

<summary>What are K-values?</summary>

The K-value is a critical metric that determines the value of a specific GPU specification within the Aethir network. It directly influences the distribution of rewards and the staking thresholds required for various GPU models.&#x20;

</details>

<details>

<summary>Why are we updating them now?</summary>

This is the first major update since launch, driven by:

* Changes in market GPU pricing
* Better aligning staking and rewards with GPU value, demand, and availability
* Preparation for upcoming feature rollouts and broader hardware support

</details>

<details>

<summary>Who is affected by this change?</summary>

Only new GPUs are subject to the updated staking amount, but reward changes apply to all GPUs.

</details>

<details>

<summary>How does the new reward structure incentivize higher performance?</summary>

* If you maintain high uptime and utilization, your relative rewards likely increase
* If you provide consistent service, you benefit from the rebalancing
* Total reward pools are growing with network usage, so the pie is getting bigger while distribution becomes more merit-based
* Some lower-tier GPUs may receive slightly adjusted rewards (a detailed breakdown is shared in the official documentation)

</details>

<details>

<summary>Do I need to do anything?</summary>

**No**. The changes are applied automatically:

* No manual claim or staking adjustment is needed
* Your dashboard will reflect updated staking requirements for adding new nodes

</details>

<details>

<summary>Where can I see the new K-values?</summary>

You can view the updated K-values and reward calculations here: <https://docs.aethir.com/aethir-cloud/aethir-cloud-host/operational-requirements-for-cloud-hosts/staking-as-cloud-host/staking-parameters>

</details>

<details>

<summary>How often will K-values be updated going forward?</summary>

We plan to review K-values periodically based on market conditions, network performance, and ecosystem feedback — no timeline for future K-value changes is currently set.

</details>

<details>

<summary>I just staked recently — will my rewards be recalculated?</summary>

Yes. Rewards will be affected for both new and existing GPUs.

</details>

<details>

<summary>What if I unstake and restake the same GPU?</summary>

Unstaking and restaking after the update will trigger the new K-value for staking. If you want to retain your current terms, it’s best to leave existing nodes staked.

</details>

<details>

<summary>How were the new K-values and rewards calculated?</summary>

They are based on a combination of:

* GPU market prices
* Supply/demand dynamics in the network
* Performance benchmarks across workloads

</details>

<details>

<summary>How does this help the network?</summary>

It ensures:

* Long-term sustainability of rewards
* Fair treatment across Cloud Host sizes and GPU specifications
* A more balanced and stable network as new GPU specifications join

</details>

<details>

<summary>How does adjusting K-values drive AI customer demand?</summary>

Simple economics:

* Adjusted K-values → Lower staking requirements
* Lower staking requirements → More GPU supply&#x20;
* More GPU supply → Happier AI and Gaming clients
* Happier clients → Maximum utilization for Cloud Hosts
* Maximum utilization → Higher rewards

</details>

<details>

<summary>Why does capturing AI demand matter for Cloud Hosts?</summary>

Because demand drives utilization, and utilization drives rewards:

* Network with the most AI clients = highest GPU utilization
* Highest utilization = maximum revenue per GPU
* It's better to be on THE platform than A platform
* Network effects mean winners takes most

</details>

<details>

<summary>How does this make Aethir #1 for GPU-as-a-Service?</summary>

By solving both sides of the market:

* **For AI builders:** Best prices, instant availability, unlimited scale
* **For Cloud Hosts:** Maximum utilization, consistent demand, best returns
* When both sides prefer Aethir, we become the default AI infrastructure network

</details>

<details>

<summary>Does this mean Aethir isn't making money?</summary>

The opposite. We're adjusting K-values because:

* Network demand is strong and growing
* We can optimize for utilization with lower barriers to entry
* The market opportunity justifies further expansion
* We're confident in sustained demand growth - Companies in trouble raise barriers; successful companies lower them to scale

</details>

<details>

<summary>What about providers who staked at higher K-values?</summary>

Existing stake is unaffected. You maintain:

* Your current staking terms
* Your position in the network
* Any locked rewards or vesting schedules

</details>

<details>

<summary>How does this affect different GPU types?</summary>

Each GPU specification sees proportional K-value adjustment, but the impact varies:

* Lower-end GPUs: Become more accessible for Cloud Hosts to contribute
* Higher-end GPUs: Maintain premium position, with lower entry barriers
* The relative value hierarchy remains, but absolute staking requirements drop

</details>

<details>

<summary>Why not just increase rewards instead of lowering K-values?</summary>

K-values do more than set reward levels—they determine:

* Network accessibility and growth potential
* How rewards are distributed across different GPU specifications
* The balance between stake and performance
* Market positioning versus competitors - Simply increasing rewards doesn't achieve these strategic objectives

</details>

<details>

<summary>How confident are you about AI demand growth?</summary>

Extremely confident based on market signals:

* Every tech company is pivoting to AI
* GPU shortages are getting worse, not better
* AI model sizes are growing exponentially
* Inference workloads exploding
* No signs of slowing down - the question isn't if demand will grow, but how fast

</details>

<details>

<summary>Will major clouds just lower their prices to compete?</summary>

They can't match our economics:

* They have massive fixed costs and overhead
* Shareholders demand high margins
* The centralized model can't scale efficiently
* They're supply-constrained - we can sustain prices they can't touch

</details>

<details>

<summary>Is this a one-time change?</summary>

K-values will be reviewed periodically as the network evolves, but changes will be:

* Infrequent and data-driven
* Announced transparently
* Based on network performance metrics
* Designed to maintain stability while enabling growth

</details>

<details>

<summary>What if this attracts low-quality providers?</summary>

Our quality controls remain unchanged:

* Performance requirements: Still stringent
* KYC requirements: Still mandatory
* Slashing mechanisms: Still active
* Monitoring systems: Still enforcing standards

</details>

<details>

<summary>Why should AI builders trust Aethir?</summary>

Because we solve their biggest pain points:

* No more begging for GPU allocations
* No more runway-killing prices
* No more capacity constraints
* Just GPUs when and where they need them, at prices that work

</details>

<details>

<summary>How quickly will we see increased demand?</summary>

The progression will be rapid:

* Immediate: New Cloud Hosts and additional GPUs onboarding
* Weeks: Capacity expansion visible to the market
* Months: AI builders migrating to Aethir
* Quarters: Utilization rates climbing significantly
* Years: Aethir recognized as the GPU compute leader

</details>

<details>

<summary>Bottom line: Should I be concerned?</summary>

Only if you're not providing quality service. This change:

* Rewards active, reliable Cloud Hosts
* Positions Aethir to capture massive AI demand
* Accelerates network growth and utilization
* Benefits everyone positioned for the AI boom

**The Reality**: K-values were conservative protective measures. Now that we've proven the model works and AI demand is exploding, we're optimizing for growth. This is what winning looks like.

<br>

</details>