Nearly every tech company seems to be working on AI tooling, and everyone is at least talking about using it. But how the heck are companies going to charge for it?
There’s little clarity on the business model front, which contrasts with the clear optimism on the part of corporate leaders and their customers about the potential impact of new AI technology on their products and markets.
The Exchange explores startups, markets and money.
Read it every morning on TechCrunch+ or get The Exchange newsletter every Saturday.
Regular readers of The Exchange will recall that we touched on this very question in May. At the time, we took a broader view, considering how consumer-focused AI services and AI-powered business services would charge.
This morning, we’re narrowing our focus to the enterprise software space, the target of an enormous number of startups. Let’s start by hearing what the CEOs of Salesforce, Box and CrowdStrike have to say on their AI-related work and customer interest. Then, we’ll go over their answers to industry analysts regarding how they intend to charge for new AI products.
We’ll wrap with a few thoughts on where startups might have an edge and where they probably won’t.
If you build it, they will come
Box, Salesforce and CrowdStrike are very different from one another. Box focuses on enterprise storage and productivity, Salesforce is a CRM giant and CrowdStrike is a cybersecurity company. However, they share a very interesting facet: They store or create lots of data for their customers.
Box, of course, started as an enterprise file storage and sync (EFSS) company, though its remit has expanded with time. Salesforce has oodles of its customer and sales data, in addition to a variety of other information. CrowdStrike has an enormous historical archive of cybersecurity-related data that it uses to help predict and interdict new cyber threats.
Data is the unifying factor here, and since data is what today’s AI models love, it’s not surprising that the three companies had a lot to say on their earnings calls regarding how they will use, or are using, AI.
Here’s how Salesforce CEO Marc Benioff described his company’s perspective on new AI technologies’ impact on business software and technology. Note his comments on how corporate execs are quite curious (condensed, emphasis ours):
The coming wave of generative AI will be more revolutionary than any technology innovation that’s come before in our lifetime, or maybe any lifetime. Like Netscape Navigator, which opened the door to a greater internet, a new door has opened with generative AI.
And it is reshaping our world in ways that we’ve never imagined. Every CEO realizes they’re going to have to invest in AI aggressively to remain competitive, and Salesforce is going to be their trusted partner to get them to do just that. Every CEO I’ve spoken with sees AI as a revolution, beginning and ending with the customer. And every CIO I’ve spoken with wants more productivity, more automation, and more intelligence through using AI.
Here how Box CEO Aaron Levie described his company’s perspective on why proprietary data matters for generative AI tooling (condensed, emphasis ours):
We are at the beginning stages of a new era of software. […] As highlighted by the meteoric rise of ChatGPT, we’ve recently begun to see a huge breakthrough in the potential of Large Language Models, or LLMs, which are now capable of bringing human-level reasoning to a large number of tasks. However, the real power of these new AI models is when you use their intelligence to help you work securely with your own proprietary dataset.
[Historically, we’ve] had limited ability to ask questions of our unstructured data, like content, which is 80% of corporate data. And now we can. By safely bringing leading AI models to enterprise data, enterprises can truly unlock the value that lies in their content.
To do this, we need a way to connect these models safely, securely and compliantly to our enterprise content. As we announced just earlier this month, with Box AI, we’re taking the power of the world’s leading AI models […] With Box AI, customers can ask questions of their content or generate new information leveraging Box Notes.
But this is just the beginning. Ultimately, as a core platform capability, Box AI will be used throughout the product to continue to transform how we work with our content in a variety of ways.
Finally, here’s how CrowdStrike CEO George Kurtz described his company’s perspective on AI tech and the importance of proprietary data in their training (condensed, emphasis ours):
For years, our use of AI has enabled us to rapidly scale that business to a leadership position with an exceptional product margin that exceeds our overall company gross margin. […] While others are just now jumping on the AI bandwagon, we have transformed cybersecurity with an AI-powered cloud business from inception. Generative AI is transforming the world, and security is no exception. Large language models, or LLMs, are only as good as the data on which they are trained, and human-annotated content makes for the best training data.
Driving better customer outcomes relies on having a data advantage and the context derived from that data. While we expect that LLMs will become commoditized over time, the data on which they are trained will not. CrowdStrike is uniquely positioned to benefit from this new technology. Our dataset spans petabytes and captures trillions of new events daily from our global fleet of sensors. […]
Our data advantage creates a unique competitive moat, yields better models, better automation, and better outcomes. We see the rapidly growing adoption of generative AI as a democratizing force within cybersecurity from both an adversarial and protection standpoint.
With all of this in mind, we introduced Charlotte AI, an exciting new generative AI security analyst utilizing CrowdStrike’s high-fidelity data advantage. […] Charlotte AI represents CrowdStrike’s latest innovation in helping security teams worldwide contend with the cybersecurity skills gap, respond to threats faster, and reduce operational costs.
Summarizing: Everyone wants to wring productivity gains out of AI, and companies that have lots of data are bullish they can do something special with AI.
Show me the money
The three quotes above allow us to make a small prediction: The bigger the dataset an enterprise software company has, and the more critical and private that data is, the greater the chances that it will be able to do something unique, and therefore valuable, with new AI technologies.
How valuable? And how will these companies charge for that value? That’s a good question. Either these companies’ leaders do not know, or they aren’t sharing.
During the same call, an analyst asked Salesforce about “the opportunity to monetize AI within your product base.” Benioff did not answer that question, and instead made an extended comment on the importance of AI, recounting a discussion with a friend-customer, and closing with the following bit of excitement:
What that means for this customer and for every customer is that they have an opportunity to transform their business. And for Salesforce, that also means an opportunity to transform ourselves; and for our industry, a new super cycle, where every company will have to transform to be AI-first.
That doesn’t exactly clear things up.
Next up, Box. When asked about “the potential for monetization” of new AI technologies by an analyst, below was how CEO Levie responded. Pay close attention to the part where he sees a possible edge for the company and where he expects new tooling to become commoditized (condensed, emphasis ours):
I do want to note [that] it’s still pretty early in our overall journey with the latest wave of our Box AI efforts. Obviously, we’ve been in the space for a while and deeply understand the kind of potential in the use cases. But with [what we are] seeing with large language models as a pretty new frontier of use cases that we are unbelievably excited about, we want to make sure that we drive the right kind of product UX, the right kind of pricing and packaging, which is why we are taking a kind of measured approach as we roll this out.
The way that we think about it internally from a product and company standpoint is there are some use cases [that are] just going to be kind of fundamentally table stakes for a product like ours. Some of those table stakes might be generating content with AI. Some of them might be asking questions of some amount of data leveraging AI. And in those cases, we want to try and ensure that we can broadly make that set of capabilities available. […]
And then there are some maybe more either advanced capabilities or capabilities that have a high consumption dynamic related to them often through our platform or maybe in workflows where we will want to have incremental monetization via either our kind of platform API business or through our multiproduct suites packaging. That will be really driven by some of the use cases that we see from customers, again, especially through this design partner program.
So I’d say kind of in the coming quarter or two, stay tuned as we continue to evolve the ultimate pricing and packaging decisions. I think you’ve seen pretty similar dynamics in other — either peer companies or big tech companies as they roll out beta programs of their products, trying to figure out what’s the right kind of pricing model. We’re basically doing the same.
Summarizing Levie, not all new AI-powered software tooling will be differentiated enough that they could charge extra for them, but some stuff will be, and the industry is still sorting out which is which. This should not surprise us — the generative AI boom is only a few quarters old, not years, so we are in the early innings.
Finally, here’s how CrowdStrike’s CEO Kurtz answered the same AI monetization question (emphasis ours):
Well, it’s something that’s really foundationally built into the platform. And we believe it’s going to drive a lot of additional adoption of modules and platform usage throughout the customer base. So, we’ll start there. As it evolves over time, we’ll look to see if we’ll monetize it with specific SKUs.
But I think, first and foremost, let’s get it to the customer base, let’s iterate it, let’s leverage the data advantage that we have because, as I’ve talked about in the earnings call, we’ve got 10 years of being able to train these algorithms. And I think, as most know, it really is the human interaction that allows LLMs to shine. We’ve got, I think, a real advantage because we’ve got 10 years of attack pairing, if you will, with data and how the attacks work that can be used for training. So, we’re going to get it out to the customer base, continue to iterate it, and then I believe it’ll drive more adoption of the platform modules.
And then, we’ll see how we’ll monetize it after that from a separate SKU perspective.
Here’s how we feel about that: ¯\_(ツ)_/¯
New tech, excitement and rapid product iteration with an uncertain business model? Sounds like a space ripe for startups to disrupt, yeah?
The only issue for startups that I can see in the above is that major tech companies simply have more customer data than startups do. We’ll see in a few months if that gives them a material advantage over startups.
My cynical side wants to argue that startups should look to build in-market leverage by collecting and retaining as much customer data as is reasonable to help them train new models or to create models tailored to customers. That might be a bit of a hasty recommendation, though, given the sheer number of data breaches we hear about these days.
Regardless, when there is a lack of clarity in the market, there’s more room than usual for new business models to sprout roots and shoots. That’s good news for enterprise software startups around the world.
If the majors don’t know how to price this stuff, it means customers probably also don’t have firm expectations. That’s an attractive attack vector, right?