On 14th of March Microsoft announced retirement of Power BI Premium. It caused a lot of noise in the Power BI Community, which is very much justified. In this article I will try to explain what will change, what does it mean for you and your organization. I will also try to play a role of a devil’s advocate here and go a little bit against the one-sided arguments you may often find online. Without further ado, let’s begin.

What happened?

 

1. Power BI Premium per capacity SKUs – I am pretty sure it doesn’t require any introduction. As costly as it was, especially for smaller organizations, everyone enjoyed the predictable pricing model.
2. Fabric Capacity pay-as-you-go SKUs – New option announced on June 1st, shortly after the world heard about Fabric for the first time. Much more costly but allowing you to pause the capacity and simply not pay for it. Which is actually great for testing the capabilities. With access to Azure portal, you may provision the Fabric Capacity in minutes and start testing. For P1 equivalent, which is F64 Capacity, 1 hour of testing costs $ 11.52. If it takes 100 hours to build the Proof of Concept, it will cost you only $ 1,152, which is huge difference comparing to full price for Premium Capacity.
3. Fabric Capacity Reservation SKUs – around 6 months later, Reservation option is announced. As promised by Microsoft, the price comparing to pay-as-you-go, is much closer to Premium SKUs. Everywhere you could hear the same question: Why we need two of them? Especially that Fabric option is not really profitable. Of course we don’t need two, and we are forced to switch to new, more pricy option.

What will change?

1. Base price will increase.
2. New storage cost is introduced.
3. Power BI Report Server is not included in Fabric SKUs.

No more Power BI Report Server

Storage cost in Fabric SKUs

We covered the one point that I totally agree with voices online, now, we will cover the one that I feel is exaggerated. I hope I get it right, and I am happy to hear from you that I am saying something stupid here. But my understanding is that the Storage Cost is associated only with OneLake. This is very important to clarify, because we often see the examples online saying, that you would have to pay over 2k USD for 100 TB storage with new SKUs. Even though the calculation is correct, this example is not the best one. It suggests that we will pay for any type of storage, as 100 TB is our total entitlement for P1 Premium Capacity. If you are using Fabric today already, in Capacity Utilization App you may find Storage page, with the summary of billable vs total storage. Today, if we want to have a Data Lake Storage in Azure, we must create a Storage Account and pay for it almost the same amount of money. So, why Data Lake Storage should be free in Fabric? It is important point to consider, when you are going to discuss this topic with your organizations.

So, how much more will you pay for Fabric because of the storage fee? Depends on if you use Lakehouses / Warehouses within Fabric. It can be almost nothing, it can be even much more than mentioned 2k USD. After all, Data Lakes are used to store vast amounts of data.

Price will be even higher if we enable Disaster Recovery. On top, we must remember, that there is a retention period defined for Power BI / Fabric workspaces. The setup is done on a tenant level, and you may choose any value between 7 and 90 days. It is important from the storage perspective. If your retention period is set to 90 days and you remove entire workspace with Data Lake storage charged against it, you will still pay for it within those 90 days period.

But again, there is more. So far, Power BI Admins were completely abstracted from complexities related to cloud services. There is one more part of the new billing that is yet to be published – Networking. There will be additional cost related to data transfer, between regions, zones, etc. For now, it’s just mentioned that this is coming, but nothing that we could cover today.

Base price increase

There is one last point to cover – the one that in my opinion is not treated seriously enough. Here is usually where people say – you will pay more, not much, but more. And here is where I strongly disagree. When we compare the prices in the screenshot above, they are almost the same. However, above both of them, we see “From”, suggesting, that this is the price for lowest configuration.

One important point to mention here, is that Fabric has smaller SKUs, but since this is Power BI Premium Pricing page, it starts with the F64 Capacity, which is the equivalent of P1 SKU. But of course, in both cases (P and F SKUs) we could have more powerful setup, with higher price. That’s what explains the “From”. Or is it?

Everything gets more complicated when we enter the Azure realm. The price for F64 that is provided in pricing page is for the US Region, and it is the lowest one we can get. Once we move outside of the US, the price is higher. For example, UK Region Capacity cost is 5.8k USD, for West Europe it is 6.1k USD. For West Europe this is already 12k USD more on a yearly basis per Capacity.

This is no longer “slightly more”. Unfortunately, that’s not all. So far, Power BI Licensing could be a subject to significant discounts, negotiated together with all Office 365 and Power Platform licenses. With Azure Services these discounts are usually much lower. If we consider realistic example of 20% discount for Premium Capacity in West Europe, vs 5% discount for Fabric Capacity in the same region we get 0.95 * 6.1 – 0.8 * 5 = 1.8k USD. This could be the monthly difference on a single West Europe Capacity, meaning over 21k USD yearly.

Laundry truck escape

In “Naked Gun 33 1/3”, Frank Drebin planned to escape from prison by digging the tunnel. As he said “It was either that or the laundry truck. But the thought of lying nose down in underwear with skid marks just didn’t leave a good taste in my mouth.”

We knew that P SKUs will disappear sooner or later. However, for whatever reason, Microsoft decided to take us there using the laundry truck, and there are couple of skid marks that will leave a bad taste in our mouths:

1. More expensive solution– as acceptable as it may be (everything gets more expensive around the world), it will be a hard thing to swallow at the beginning for many organizations.
2. Charging more for less– of course, with Fabric you gain a lot, but what about organizations that don’t use it? Seems that Microsoft forgot about them.
3. It happened too soon– General Availability of Fabric was announced too soon in my opinion. It becomes a new standard, that documentation for each Fabric workload has a long of list limitations and there is a lot of features that are very important for customers, available in existing tools (like Azure Data Factory). This rather leads to the conclusion, that GA of Fabric was rushed, and product is not really finalized. Is it ok to start charging more already?
4. Disruption of status quo– considering Microsoft standards when it comes to licensing models, Power BI Users were probably the luckiest. Everyone already got used to frequent changes in the tool, but at least we didn’t have to bother with complex licensing, and how to keep the cost at the lowest possible level. That time is over. Organizations will be forced to rethink their architecture, check options to move capacities to cheaper Regions, etc.
5. Stakes are too high– There is so much to lose here. Microsoft entered the Data Platform game a bit too late. They had not a very good start with Azure Synapse, now they are trying with Fabric. It happens when competitors have already established their position in the market. Microsoft spent almost entire year promoting Fabric, rushing the General Availability of the tool to convince people that everything is hunky-dory, only to make people stop for a moment, and re-think their strategy. I don’t get how important it was to drop the bomb already, and risk losing customers at this stage.
6. Not exactly a reward that was expected– There are customers supporting Power BI since the very beginning of the tool. Using it on early stage, constantly helping to improve it, providing feedback, and basically paying for growth by using the tool. Amazing community helping each other, convincing new customers to go with Power BI as their analytics platform. Thanks to combined effort of all these people, Microsoft managed to secure their position as leaders in Gartner Magic Quadrant for Analytics & BI Platforms. I really hope that this is not the way for Microsoft telling them “Thank you for your service”, and there is still a room for negotiation and keeping healthy relationship.

Devil’s advocate

As promised, I will try to defend Microsoft a bit. It actually already started when I was talking about Storage cost, that is introduced in Fabrick SKUs. For now, Microsoft is giving it away for free, as a pat of Power BI Premium. So, it’s not like we must suddenly pay for something new, but we simply must pay for Data Lake Storage, which we haven’t had to pay for until now.

Regarding price increase, we had a pretty good run with current pricing as the 5k USD price was basically frozen for years. Organizations needed to increase their prices many times during these years, due to growing costs, lack of resources, etc. Should we expect then that Microsoft’s price will stay forever on the same level? Probably not. Besides, even if Microsoft solution will now be more expensive, is it more expensive than the offer from competition? I also hope, that with this huge change, we will again have a couple of peaceful years ahead of us, with prices staying on the same level.

Discover new opportunities. If you didn’t consider using Fabric, and from your point of view you will be forced to pay for Fabric even though you don’t use it, maybe it’s good time to start looking into the value that Fabric Platform could bring to your organization? Maybe you will pay more for Fabric Capacities, but you will save a lot on existing cloud infrastructure? If you were using Azure Synapse Analytics and Azure Data Factory, you know that both of them come with significant cost. Microsoft Fabric allows you to manage your workloads in much more efficient way. Helping you to fit in the SKU size of your choice.

It is often brought in the discussion that Power BI Premium per User is still an option. I don’t like this argument as it’s kind of rude thing to say. While it is important to know all your options, I am much more in favor of argument that I used above – if you are forced to pay more, try to find new value in the Fabric Data Platform.

Finally, the good thing about this change is that there is a safety net for large organizations, having Enterprise Agreement with Microsoft. It is said that nothing will change for you till your EA is still valid. This gives some time to prepare for change and start serious negotiation with Microsoft. Since this is the case, I like to think that not all is lost yet. Maybe I am naive, but I want to believe that Microsoft will honor the long-term relationships with their customers and will meet their expectations half-way.

Just to make things clear, I am not justifying Microsoft’s decision. I am trying to understand what happened, using my limited knowledge. I always prefer the debate instead of one-sided bashing, and I hope to see it regarding Premium SKUs retirement.

My own stand

As a person who fell in love in Power BI, and feels the same about Fabric, my goal can be only one. I really want to see how this platform succeeds. I see a lot of value in Fabric, and I want to see it growing into something even greater. Even if Microsoft is not making it easier, I will continue spreading a good work about Fabric, and, most importantly, Power BI.

Conclusion

Without any doubt, this is going to be a hot season for Premium users. If I were at any position to advice, I would advise to stay calm and don’t jump into different platforms. Let’s grab a phone instead and try to talk to MS and negotiate a better offer. As naive as it may be, I really hope this is what happens, and Fabric as a unified Data Platform will grow.

As always, thank you for reading, and see you in next article.