Cloud Pricing Strategies | Founders' Review Editorial

In the ever-evolving landscape of technology, cloud computing has emerged as a core element, offering scalable, efficient, and flexible solutions for businesses and individuals. At the core of utilizing cloud services effectively lies the understanding of cloud pricing strategies. This article dives into the various pricing models adopted by cloud service providers, offering insights into how businesses can navigate and optimize their cloud expenditures.

Pay-As-You-Go / Serverless / On Demand - It was the original cloud pricing strategy to invite new customers. Serverless services are good for starters. Free trials are the entry for serverless pricing. You just tell one simple task to do, and the system covers that for you. Later, you can switch to Pay-As-You go as your cash flow grows. It is not just affordable to pay by the service use. You do not need to set up the hardware and operating systems. Serverless is ideal for startups and small businesses. Pay-As-You-Go is priced as a utility bill by metered usage of service calls or container use time. Serverless is the most common use case but you can do the same with bare metal and virtual instances. This pricing strategy is simple for funded cloud startups. You buy a datacenter in advance. If 50-60% of the resources are paid as Pay-As-You-Go, then you offer the rest to new customers for experimentation. The idea behind Serverless pricing is the following. 10% Serverless utilization can pay for the entire 100% capacity of the cloud provider side. Serverless is usually priced 10x in exchange for low latency. Flexibility is the key. Growing into the full capacity only when needed like in case of a sports event is very useful. This is how the big three AWS, Azure, and Google Cloud were built up.

Subscription / Reservation - Reserved Capacity can be used for entire virtual machine instances. These can be configured just like on premises clusters. Virtual machines save on hardware staff. You can run as many services on them as you want. There is no need to negotiate with management. Utilization is low as a result, but billing is simple. Subscriptions and reserved capacity are usually a good choice, when organizational budgets and limits are set in advance. Long-running services are typical to run on reserved instances. Reserved capacity gives a better price for the infrastructure services usually for a half or one third of Pay-As-You-Go or one tenth of Serverless pricing, in exchange for a commitment to pay upfront, or pay monthly for an extended period of one to three years. Some reservations can be resold, if the capacity is underutilized. This makes them ideal for big budget high demand enterprise workloads. Fierce competition in mature SaaS service markets trigger the use of Subscriptions / Reservations, when price is the key for further growth.

Spot Pricing - Sometimes tasks can be delayed. Such workloads can use spot pricing strategy. Opportunistic Spot prices are usually fluctuating. This is followed by fluctuating spot demand as well. Such instances are preemptible as a result meaning that they can be terminated by the cloud provider without notice. The cloud provider gives these spot instances, when higher paying reserved or serverless users request capacity. Spot pricing can be one fifth or one tenth of the standard Pay-As-You-Go pricing, but the total cost may vary, since significant computed results may be lost at termination. Spot pricing led to new innovation like Spark Streaming and Apache Flink.

Token Pricing - Cloud pricing was always the culprit of growth. Budgets are tight. Rates are low but only, if you can match them with high quality, stable revenue workloads. Workloads may have bugs consuming more resources than planned. Demand may be unpredictable. Cloud spending is a major issue triggering many meetings allocating management time. This led to less growth in cloud provider revenues. A better approach is token pricing to spark further growth in cloud computing. This is a modern approach replacing cash with tokens our coupons like computer games. OpenAI uses this method in their SaaS offerings.

Whether Token Pricing is the future depends on many things. The United States Securities and Exchange Commission approved BitCoin based spot ETFs for the first time. Distributed cryptocurrencies like BitCoin technology are a vehicle for investment bank and central bank use due to their mathematical complexity. It is less supportive of retail use. It is just too complicated compared to how green is your cash. Their popularity probably grew as the technology was so power hungry that you could use it for heating in the cold parts of Russia. There were many more cryptocurrency technologies invented since then. Some of them are less power hungry, and BitCoin received low power updates and shims, shaving off of the original idea of legal stability.

Simple liability based coupons, vouchers, or tokens are probably the way to handle organizational budgets across cloud providers in the future. The reason is the simple solution to reduce risks securing dividend payments of bonds used for long-term assets.

Token pricing uses prepaid tokens to pay for services. Token spending is still set by the customer budget. However, tokens are used to purchase reservations, subscriptions, spot items, or Pay-As-You-Go serverless services. This allows fluctuation of the actual pricing reducing risks of the cloud providers.

Cash payments are hard to communicate, manage and account. Cash budget and payment is allocated at the beginning of the period in exchange for cloud tokens. The token based approach simplifies the payment process.

Teams can still deal with budgets, but once the budget is set, they can be more flexible with usage. Tokens are fungible and they can be sliced and diced among team services.

The tokens allow fulfilling a flexible demand with inflexible supply. They do so by changing the supply, the amount of services given for the prepaid tokens. Services that are bound to serverless or spot can be postponed to the next period.

Tokens help the cloud provider. Tokens are bought in advance. Any hardware can be financed using the original payment removing significant risks from the system. The planned usage can be estimated better with pre-paid tokens. Underutilization can be handled by changing the offer for each token increasing the value customers get.

Tokens are worry free. Spikes in demand used to cause service disruption and dropped connections. Serverless services allowed fulfilling the extra demand with extra unexpected spending.

Excess spending above the long term average can offset into discount tokens. These discount tokens spread the extra usage across a time period. This supports both predictability and cloud provider growth. Customers will more likely opt for serverless as a result.

A customer may need to pay a few thousand dollars for a spike. They get a discount in the near term fixed costs making the extra payment disappear. This approach saves on management, negotiations, and risks. The discount leverages the utilization gap of the serverless cluster, so the cloud provider gains a stable revenue and a loyal customer in exchange.

Discount tokens are way more simple than providing the same elasticity with payments. There is lots of accounting, management, and sales overhead. Real money movement is per period. Granular, fungible tokens are just a liability in the cloud provider's system.

More advanced reserve bank style strategies can help to reduce cloud costs. Cloud costs can be covered by floating value tokens like coupons. Tokens are bought prepaid, and they have a term tied to the lifetime of the hardware bought at the time of the purchase. This removes significant risk from the cloud provider's business.

Are there more possibilities to reduce risks? Probably yes, but our approach covers the major sources of variance in cloud demand and supply. These strategies should be good starting points to start reducing risks for business professionals.