Generative AI: An Asteroid Impact Event on B2B Software
Speed and Adaptability Will Define the B2B SaaS Winners in the Generative AI (R)evolution
To say there’s no shortage of opinions on ChatGPT and the future of AI would be an understatement. With voices ranging from AI researchers, builders, and investors to the notorious clickbait “AI influencer” on social media, it seems like everyone has something to say. Perhaps I shouldn’t add to the noise … but it’s hard to keep up with the pace at which AI has been advancing in recent weeks and months, and putting my thoughts together in a blog post helps me sort them out (and often leads to great feedback and discussions).
I got my first computer (a Commodore 128D) in third grade (around 1986, which makes me feel very old). The recent advancements in generative AI, including ChatGPT, Midjourney, and the many products and experiments built upon OpenAI’s GPT-3/4 and other APIs, have blown my mind as much as my first experiences telling the C128D what to do in BASIC, playing games on the Amiga a few years later, and exploring the WWW in the mid-90s. I’m sure some people will think I’ve drunk too much Kool-Aid, but I’m genuinely convinced that generative AI will cause an asteroid impact level hit in the world of B2B SaaS.
That leads to numerous questions, e.g. to which extent the playbooks on how to build, fund, and scale SaaS businesses will have to be rewritten. Louis has been writing about these topics for years (e.g. here, here, here, here, and yesterday here) and will continue to do so in the light of the latest developments. In this post, I’d like to focus on one question only:
Is AI a platform shift as disruptive as the move from on-premise to the Cloud?
As the world moved to the Cloud in the 2000s and 2010s, a large percentage of the value was created by new SaaS startups, not by on-premise incumbents who managed to move to the Cloud. Intuit, Adobe, Oracle, and a few other older software companies have managed to transition to a subscription-based business model. But hundreds of billions of dollars in market cap have been created by Salesforce, ServiceNow, Hubspot, Workday, Atlassian, Veeva, and many other companies that were started after 1999.
Moving from on-premise to the Cloud required a complete rebuild. There is no smooth transition path from single-tenant databases and desktop UIs to scalable, multi-tenant databases and a UI that runs in the browser. It’s a complete overhaul of the software architecture, rebuilding it from the ground up. Coupled with a business model change that led to reduced cash flow in the short term and the massive changes in sales & marketing that came with SaaS, it’s easy to see why this was too much for most software companies that grew up in the 80s and 90s.
Is the same about to happen with AI? Are most SaaS companies going to be replaced by new AI-first startups? So far, it appears like today’s SaaS leaders are in a much better position to embrace AI than on-premise software companies were to move to the Cloud. Here are a few reasons why:
1) While ChatGPT was the iPhone moment for AI that catapulted the topic to the top of everyone’s priority list, AI is, of course, not new. Deep learning has been making steady progress over the last decade, and AI has been used extensively in a large variety of consumer applications. So AI has been on the agenda of SaaS leaders for some time. Salesforce, for example, has acquired several AI startups starting in 2014, and Zendesk released an ML-powered chatbot in 2016.
2) Developing AI features doesn’t require a complete rebuild. About three years ago, it started to become much easier to add AI with a few open API calls, e.g. by using GPT-3’s API. Today there’s a large and fast-growing ecosystem of developer tools that help SaaS companies leverage the power of LLMs. So using AI doesn’t mean you have to throw everything away. The opposite is true — SaaS leaders can leverage their existing customers, data, and expertise to build things that would be much harder to do for new entrants. Hubspot, Intercom, Notion, and others have moved into generative AI at impressive speed, showing that if you can move fast (not a given at all for companies with hundreds of millions or billions in ARR), you’re in a good spot.
The counter-argument is that “slapping AI onto an existing SaaS app” is not the way to go, and I’m sure that in many cases, that will turn out to be true. I have no doubts that building and using software will change more radically than most people can imagine today. And maybe that means that if we look back in ten years, a large part of the value creation in B2B software will have come from new AI software startups started in the 2020s. Given the pace of AI development in the last few months, it’s hard to predict anything. Maybe we’re all replaced by paper clips sooner than we think. ;-)
But for now, my view is that SaaS companies have a good chance to survive the shift to AI and come out of it even stronger — delivering more value to customers and doing it with a lower cost base by leveraging AI to increase productivity — IF they go all-in on AI now. That is a big IF, though.
Adapt or perish
Large species, like humans or elephants or dinosaurs, with lifespans in years or decades, are slow to adapt. If living conditions suddenly change — e.g., because of rapid climate change or an asteroid impact — small species with short individual lifespans and quick reproductive cycles tend to have much better chances to adapt fast enough and fill ecological niches that have opened up.
Similarly, companies tend to lose the ability to change direction rapidly as they get bigger and bigger. In the case of Intercom, the company was able to move fast because the founders made it a top priority, and once their ML team had an initial play with ChatGPT, heavily resourced it. In the case of Hubspot, Dharmesh wrote ChatSpot.ai with a tiny team to move at startup speed without distracting the core R&D team. I’m stretching the analogy with evolution here, but you could argue that Intercom managed to leverage cultural adaptation (which is much faster than genetic adaptation), while Hubspot increased genetic diversity (which pushes adaptation) by creating a Hubspot offspring called ChatSpot.ai.
When an asteroid hits, only the species most responsive to change will survive. Generative AI is an asteroid impact event for B2B software, but unlike most on-premise software companies in the 2000s, the SaaS leaders of today aren’t doomed. It all depends on how fast they can adapt.
PS: I’ll moderate a discussion about the impact of Generative AI on SaaS at the SaaStr Europa conference in London next month. Hope to see you there!
Thank you Louis Coppey, Ricardo Sequerra Amram, Aleksandra Zorylo, Tilman Langer, and ChatGPT, for your feedback and help!
