By Amazon’s own estimates, SageMaker, its AWS-based platform for machine learning workflows, has gained tens of thousands of users since it launched in 2017.
But while its popularity — and AWS’ reputation as a best-in-class cloud services provider — can make it feel like the logical choice for building your machine learning product, the truth is that it also comes with a significant downside.
One that might derail your project before it’s even out of the gate.
Here’s why you should approach SageMaker and AWS with caution, and a look at what makes iSquare, our AI cloud services platform, a more flexible and, well, better choice.
Peeking under AWS’ hood
SageMaker’s reputation as a top notch platform is more than just hype.
From basic model building and experimentation, to training, deployment, and management, it puts a comprehensive set of tools at your disposal that allow you to do everything in one place. This includes Jupyter notebooks that link and upload directly to Github, pre-configured environments, plus access to state of the art GPUs.
Of course, SageMaker is also integrated into the wider AWS ecosystem, which means you have easy access to more than 70 additional products and services. But the kicker is that Amazon offers a wide range of free trials, discounted introductory rates, and other deals.
The upside is that, if you want to build, deploy and scale a machine learning product quickly, SageMaker gives you access to a plethora of advanced, high-powered tools at minimal upfront cost.
The downside is that free trials, discounts, and deals eventually end.
At which point your costs can spiral out of control faster than you can say AI.
An insidiously sticky proposition
Whether your product is in the building, training, or deployment phase, chances are SageMaker or the broader AWS ecosystem has the tools you need to fix any issues that come up, plus countless enhancements and improvements you can add on top.
This is great, until you realise how easy it is to get caught up and start trying one tool after another, even when those tools are far more powerful and advanced than what you actually need. And because they’re discounted or have free trials, you may not consider the costs before adding them to your stack.
Needless to say, once those free trials or discounts end, you may find you’re in a pickle.
On the one hand, you’re on the hook for tens of thousands of dollars’ worth in subscription fees for hardware and software you might not even be using. Fees you might not be able to afford because you’re still an early-stage startup, or your product isn’t profitable yet.
But, on the other hand, you’re locked in.
You and your team have invested hundreds of hours learning your way around SageMaker and the AWS environment. And your product may depend on tools and services that are easiest to access on AWS.
As a result, migrating to a different provider may be equally unfeasible. Aside from the potentially hefty cost of the migration itself, you may also have to make fundamental changes to the product you’ve just spent months building.
Introducing: the simpler, more cost-effective, and efficient way to deploy and scale AI
If SageMaker and the AWS ecosystem risk sucking you in without you realising until it’s too late, the good news is that there’s a better alternative. We’ve built an AI cloud services platform, called iSquare, that lets you deploy machine learning products fast, without the sticker shock.
iSquare has a pay-per-use pricing model and no upfront setup costs, so you only ever pay for what you’re actually using. No more, no less.
More to the point, the platform is designed to automatically choose the most cost-effective and efficient hardware and server setup for you. Which means there’s no risk of you accidentally getting locked into services that give you more power and bandwidth than you need.
Better still, because it’s serverless, you don’t need to hire DevOps to manage the deployment. Upload your model, and iSquare will build, test, benchmark and deploy it on the correct hardware, produce an API you can integrate and run anywhere, and take care of your load balancing as you scale. And if you don’t have an AI model, we have a catalogue of ready-made ones to start you off. .
With iSquare, you can go from model to launch in a matter of minutes, where it takes 60% of companies up to 12 months to deploy.
Don’t get caught out
SageMaker — and the AWS ecosystem — may be packed with attractive features. But while this makes it hard to resist its pull, it’s also a big risk for your business.
As a user on the ycombinator forum puts it: ‘The scariest thing… about AWS is that I might accidentally bankrupt myself while I learn to use it.‘
And it’s not just those who are overenthusiastic about learning the ropes who risk getting slapped with unpleasantly hefty surprise bills.
When social media giant Pinterest had an unexpected spike in traffic in 2018, for instance, they racked up a $20 million bill on top of the $190 million they’d paid AWS in advance.
At Alpine Intuition, we’ve created a simpler, more cost-effective way to deploy machine learning tools in a fraction of the time it takes your competitors to do it.
Because, if we make AI development more accessible to more people, who knows what exciting new opportunities the future will bring?
Want to find out?
Let’s talk about how we can help you deploy AI at scale in minutes, not months
About Alpine Intuition
We believe that Artificial Intelligence is going to play a major role in the future. Therefore, we think that it is important that the understanding of the possibilities, and the potential applications of AI are made widely available. At Alpine Intuition we set the goal of democratizing AI through our products and services.
For more information about our company and solutions, please visit alpineintuition.com
For press enquiries, please email: [email protected]