How much of a differentiator is being AI-native for new B2B startups?

Or in other words, who will win the race to AI – startups or incumbents?

A whole new crop of startups have sprung up as a result of the AI boom–with website headlines full of "AI co-pilots," "conversational X," or "chatGPT for Y." Some are creating entirely new product categories, because of what generative AI enables (e.g., legal tech, video generation). Others are taking on existing categories with the differentiation of being "AI-native." 

Having worked as a product marketer, I've realized just how important having a strong market differentiation is, in building a successful company. This may seem obvious, but many founders are unrealistic (or not paranoid enough) about their market differentiation. Remember – differentiation is in the eye of the customer. 

When evaluating if this new crop of AI startups genuinely have a competitive differentiation, I ask myself – in B2B SaaS, how big of a moat is being AI native? And what's stopping incumbents from simply enhancing their existing technologies–and more valuably–their existing data, with AI? 

Note: for the purposes of this article I’m not classifying AI research companies like OpenAI, Anthropic etc as B2B SaaS “startups” (although I’m sure they are desperately trying to figure out their B2B GTM). That being said, the absolute giant incumbents of the tech world (Microsoft, Google, Amazon) have already ensured their businesses will not be disrupted by these companies by becoming their major investors and technology partners.

Looking back: what can we learn from the cloud transition?

To explore this question I thought we could look back in software history to another important transition: the transition to cloud. A great example of a startup winner from this transition is PlanGrid. PlanGrid was the first construction digitization software on the app store. Tracy Young, co-founder and CEO has said “We nailed product market fit immediately, which I understand is rare. We were lucky.. we were in the right place at the right time” (source). 

Three big trends propelled PlanGrid forward - the rise of the cloud, the rise of mobile and the launch of the iPad. But still, Plangrid was a small startup with a $20M Series A – what was stopping a large company like Autodesk (who would later go on to acquire PlanGrid at $900M) from simply investing $10M into building Plangrid themselves? Was it:

• Belief: Did they not believe in the idea, and therefore it was too risky? 

• Prioritization: Was R&D not a top priority for the business?

• Speed: Were they too slow internally to build new products?

Whatever their reasoning, that decision cost Autodesk $900M. Fast forward 7 years, and Autodesk clearly believed in the idea, and PlanGrid had proved this was a massive market with over $100M ARR. At a 9x revenue multiple, this was a good deal, but still, an expensive one. 

Why don't incumbents always win?

There are many famous MBA case studies of incumbents not reacting fast enough or completely dismissing a new technology that would later disrupt their business: Blockbuster and streaming, Cisco and streaming, Teradata and the cloud data warehouse, etc.

This is despite the fact that incumbents have inherent advantages against startups, like their massive pool of resources, and most importantly, their existing customer base and willing sales team. 

Most of these case studies conclude that the large company never saw it coming – that they didn't have the vision and belief in the idea like their startup competitors did. 

So it's not so much a question of if they could have built it, it's that (at the time) they didn't want to.

How will the AI transition compare to the cloud transition? 

Many large companies didn't react fast enough to the cloud, or mobile, and that opened up a whole new market for cloud-native, and mobile-first startups to be born and take over existing categories. The question is, is the same revolution about to happen with AI? 

Let's compare and see what we can learn: 

Belief: Are people skeptical about the idea? 

Large companies were skeptical about the cloud because of the large amount of resources cloud migrations would need, and there were still many unanswered technical and security questions. Whilst people are skeptical about AI accuracy, use cases, and security, I feel that AI has a lot more hype around it, especially at the executive level. Many large companies are on high alert, having missed out on core opportunities from the cloud transition, and are eager to jump on trends this time around. 

Prioritization & Speed: How easy will it be to truly integrate AI into existing products and architectures?

If we learn from the cloud transition, it was very difficult for on-prem software companies to transition their entire architecture to the cloud. These were often multi-year projects and were difficult to get right, and in that time smaller cloud-native startups had grown in popularity. Due to the benefits of cloud infrastructure, the adoption of microservices architecture, and the availability of APIs and large language models (LLMs) like OpenAI, the barrier for AI implementation is a lot lower. 

The AI revolution is also being led by several large companies – chief among them Microsoft, sending large companies the message that it’s possible to do this. Many large companies have taken their lead from Microsoft and rushed to announce AI capabilities; for example, if we look at the construction industry, Procore has announced an "AI co-pilot." 

Of course, what remains to be seen is the success of these offerings and how integrated these user experiences will be, since many of these products are still POCs at this stage. 

Data: The missing element 

In the shift to AI, one key factor stands out: data. Unlike the cloud transition, where large quantities of data represented huge costs and technical debt, AI requires large quantities of domain-specific data for effective training. While some infrastructure changes will be necessary (enter the surge of LLM infra companies), this time around, large quantities of customer data are a significant advantage for incumbents in training their products.

Summary: Incumbents vs Startups in the AI transition

Using the learnings from above, let's summarize the advantages startups vs large companies will have in the AI transition.

Incumbent advantages

• Existing domain-specific data 

• Existing customer base 

• More resources

Startup advantages 

• Speed 

• Designing a new user experience and technical architecture from the ground up 

So...who will be the winners in the AI transition? 

Prediction: Incumbents will win existing software categories 

In existing software categories (where solutions exist beyond spreadsheets), I believe incumbents will continue to win. These are in established categories like CRM, marketing automation, and HR. Right now, it's hard to imagine the next AI-native salesforce gaining market domination in the next 20 years vs Salesforce simply adding (or acquiring) AI capabilities.

Startups with the grand vision of being the new AI-native X will lose – simply because being AI-native is not enough of a differentiator to the customer. The enterprise product bar is so much higher, and large companies have one significant advantage over them: domain-specific data. 

The caveat: when you actually need a radically new user experience

Where startups may have an advantage is if the integration of AI technology actually requires a complete overhaul of the user experience, and/or product architecture in an existing market category. In this case, it will be a battle of how quickly the startup can build their new product vs the benefit of the incumbent having a pool of existing data. In this instance, having large quantities of data could be a hindrance because the product needs to be fully re-architected to use it with AI – examples here are data analytics and data infrastructure companies.

Although many new startups competing in existing categories will claim that being AI-native involves a radically different user experience, most founders are being delusionally optimistic about this (Hint: being a chatbot is not enough). It is difficult to measure if the user experience is genuinely different enough up front, and customers will be the ultimate decision-makers here.

Startups will hit a growth ceiling and be acquired

Startups that intend to start in a niche area and grow into a platform will likely hit a growth ceiling and be acquired for <$100M. This could be seen as a deal for a public company, and a relative win if these startups stay extremely product-focused and remain lean. Founders should be realistic about the longevity of their startup, and consider building for acquisition (something that’s still seen as taboo). 

New software categories: Startups have a massive opportunity

Obviously, new categories mean no incumbents, and the AI opportunity is massive in industries where AI has unlocked a whole new market opportunity–for example in legal tech, healthcare, etc.

One thing to consider is that large companies could be more of a threat than usual here if the new category could utilize data they already have, since AI enables companies to productize their domain data quickly. Healthcare is an excellent example of this, because getting permission to use healthcare data takes time. Another radical example would be Docusign going into the contract writing space.

If you're a startup, I encourage you to really scrutinize if being AI-native is enough of a differentiator for you. 

Would love to know what you think, let me know your thoughts below.