Here the classical argument would be that it is all about network effects that allow leading SaaS providers a dominant position to be disrupted by blockchain solutions.

I think that they play an important role, but network effects are not sufficient in order to describe the economic value function of leading SaaS providers nor the case for disruption. This is because not every SaaS has a strong network effect. It is undoubtedly there for solutions like Dropbox, strengthened and supported by the freemium marketing model. But this cannot be the only argument (and yes I am aware of the downside of network effects consisting in over-exploiting the natural or created de facto monopoly position (compare "Why decentralization matters" from Chris Dixon @A16Z for the argument)¹.

I do not think that network effects are that decisive for the following reason. It is for the fact that it is hard to argue that the best in class SaaS solution for a particular business process did become the number one solution solely for network effects (while there might be a lock in). If you use Salesforce because your customer does, but Pipedrive would be a better solution, something is wrong.
There are other more important points: It is also the sales and deployment model made possible with the cloud. And there are economies of scales for the provider, a point with which I cross the line to the economic value of SaaS captured or perceived by the users. 

There is the design of processes and workflows itself, usability, worldwide accessibility, value and economies of scale. Scales that come into play from using one central database and a central server architecture that requires investments and skilled people. I think the last point is a key element as this explains why for many smaller companies – but also bigger companies - the arrival of SaaS was significant. In other words, here are two key points to list: the total cost of ownership (TOC) which went down and the possibility to move from CapEx to OpEx.

This observation makes SaaS solutions particularly interesting with regards to economic value and architectural decisions in contrast to the blockchain. It is because we can see that the centralized data processing solution first of all enables cost efficiencies which are attractive enough that IT departments keep chasing similar effects moving more and more processes to the cloud. Not to forget that SaaS as better version of ASP solved particular problems of on premise software for large Enterprises.

And enterprises as well as SMEs love centralized data as it works very well with needs of the organization. In fact, a larga number of collaborative work with cloud solutions come from avoiding versioning (e.g. a shared document on OneDrive) and central repositories.

What does this mean for blockchain solutions in the enterprise context?

Based on this description of the value creation I think we can derive two requirements to deploy the blockchain within the context of the enterprise - either it enables a totally new use case or needs to comply with a replacement logic. And within a replacement it must answer why a decentralized approach is better. 
The drivers for the replacement logic decision are also quite simple and evident. The new blockchain IT must not only be cheaper with decentralized data (defining something like a max cost boundary), the decentralized data application with same costs needs to enable a new layer of services that is value driving to justify the deployment.
Evaluating "decentralized" with regards to typical business processes companies use SaaS for - CRM, ERP, accounting and billing, just to name three – there seem to be little gains by a blockchain. 

I think this is the reason that Tomasz Tunguz² argued in one of the few articles available from VCs on Blockchain impact of SaaS that the intranet IT layer remains SaaS while the inter-company layer becomes blockchain backed.

But is this division of spaces as evident as it seems as enterprises either have networks in place or processes to work with their value chain partners?
How could one model or analogy be used to estimate the desired value capture when deploying blockchains for inter-company solutions?
And are there requirements that need to be met other than the use case checkbox charts?

How to think of value enabled by blockchain as inter-company issues?

In my point of view the value generation requirement by sharing data or collaborating is nothing new. It can be described with the classical trading approach to foster transactions with a market maker or marketplace that requires the elimination of information asymmetries in order to facilitate more transactions. Therefore, you can also define it as the marketplace approach. This definition is interesting as there are many SaaS solutions that justify a marketplace and many marketplaces that have an additional SaaS solution. Thus, specific cases become more compareable. 
But there is a caveat. The application of this to established B2B markets is not as simple as it seems as often the level of information for transactions is sufficient - and more information does not add more volume or a better economic outcome (the problem of some big data use case).

Still, let us see how a SaaS marketplace works and what impact blockchains could have. We have some old material elaborated for clients which I cannot share around SaaS and inter-company marketplaces, but I would like to point to two excellent posts by Point Nine Capital on this question.

The first one – called SaaS enabled marketplaces³ - implicitly illustrates the difficulties around B2B marketplace by describing the status quo and the investment thesis, which I think is as valid for Inter Enterprise Blockchain cases.

The second post - Will Blockchain(s) Eat the Marketplace Stack?⁴ - discusses the impact of Blockchains on marketplaces itself. There are two interesting key issue raised in the article:
"What is, for example, the defensibility or the long-term value of a marketplace which cannot rely on users' lock-in, and therefore has lower/no network effects?
What if because users own their own transaction history and reputation data they could switch to another platform seamlessly?"

If this is a problem for a VC, we can derive the related issues for an inter Enterprise blockchain network. Thus, analyzing the wide spread heatmaps it is no surprise that finance and energy trading are among the leading blockchain cases. But in these sectors trading platforms are always evolving (see the merger wave of Exchanges, the flash trading or dark pool platforms orgsnized by banks). 

To conclude this section: The marketplace approach/analogy is actually quite useful, because it allows to formulate the value creation questions not only with regards to a data administration scenario, but also the fungibility and the scope that needs to be reached deploying decentralized data like in the classical blockchain use cases.

A counter argument from use cases?

After the two layers that allow to discuss the necessary value capture when arguing for/against a blockchain solution in settings with existing IT became clear, it makes sense to go back to the level of use cases. Use cases concerning the trade of financial assets, finance and payment are obvious as the nature of the business fits well in above outlined arguments. Then I think it became clear that there are a number of cases where you might consider the deployment of a Blockchain, but its driver is not "decentralized or any other blockchain related feature", but the cost question. 
But what about decentralized as an architectural value, as it is way closer to the nature of the internet? In other words, is there a case where decentralized is as much an architectural choice as cloud which changes the classification of the blockchain from "a special form of a database".

This reminded me about a video, I watched in the regular activity to update/contrast our views with those of other industry players. The video in question is from Peter Levine from Andreessen Horowitz (https://www.youtube.com/watch?v=4QTAtFaIiyc) published about a year ago. He presents the theme "the end of cloud computing". The key argument of interest is that due to more and more data arriving from sensors and more computing power, decentralized computing will gain momentum ending the age of centralized cloud dominance, a fact we defined above of importance for SaaS.
I think his arguments are quite interesting and I am directly observing the beginning of these trends in some projects I'm following e.g a digital twin.

But with respect to the impact of SaaS solutions currently used, it is only fair to say that the video includes mainly new, future business cases around IoT, robotics and e.g. self-driving cars.

Thus, I fully agree with him when it comes to the merger of data and physical products and industrial solutions. Let us take the example of the digital twin. I think a real digital twin is defined by a thoughtful balance of centralized and decentralized computing and a very well-defined plan of which data needs to be elaborated locally and which data - for more value add or more needed processing - belongs in the cloud. I also see this happening with the increase of sensors and therefore the foundation layer that allows us to automate processes based on input data.

But all these examples bring up an interesting point that touches the nature of data entry in SaaS and a cornerstone of a pro blockchain argument. While I have no problem to think of the blockchain as the basic infrastructure for such data (and, in fact, many enterprises I talk to argue the same, but for costs), I think it illustrates a flaw of many use cases bashing centralized database solutions.

The flaw is the replacement thesis of the intermediary or the third party. There is a big difference if their use case relies on automated data recorded and screened by sensors or if the validation of data at the data entry point is a critical and value driving function performed by the user or intermediary.

Conclusion

It seems that intra company SaaS solutions seem to be relatively unimpacted by the Blockchain. This fact is supported by the insight that many PoCs we observed are centralized solutions masked as decentralized. So based on current information the impact on SaaS seems rather limited for the time being. Of course, APIs are always a hot topic that should work with Blockchain solutions.

As illustrated the situation for marketplaces and marketplace SaaS solutions is different as there are several scenarios.

The nice thing in tech is that there is always something new and markets are dynamic.

While today's focus may be consumer applications, the SaaS world will undergo a similar and fundamental transformation.

I think that although we cannot predict the future, the framework is in place to think about the impact.

Concluding there is one final thought. This is that also in tech there is a packing order of things. This means, it is important to identify the most impacting novel technology.
For this reason, I think it is interesting to contrast the impact of blockchain with artificial intelligence (AI).

While the details and implications are an article on its own, I argue that AI is likely to have way more impact as there are a number of process steps that can be automated in SaaS solutions with the help of AI and customers will request this once they see what time saving is available.

The game stays interesting.

Reference list:
[1] https://medium.com/@cdixon/why-decentralization-matters-5e3f79f7638e
[2] http://tomtunguz.com/saas-and-blockchain
[3] https://medium.com/point-nine-news/an-overview-of-the-growing-saas-enabled-marketplace-ecosystem-b4f5314356e5
[4] https://medium.com/point-nine-news/will-blockchain-s-eat-the-marketplace-stack-cf5952889aa0