Sumanth Raghavendra is one of the few bloggers in the Indian startup community who writes authentic posts, so I was naturally curious to read his latest post featured on the SaaSBoomi blog about the coming “AI apocalypse”. Of late, I had been watching from far how the SaaSBoomi community has been talking about AI and exhorting Indian founders to embrace it. I have always been a skeptic of trends (I have met more than my share of tech teams reeling with debt related to falling for the fad of the season) so I was interested in knowing what was the thesis behind this.
The post begins with a warning. Raghavendra claims that SaaS is dead. Why?
But the sad reality is that, as an industry, SaaS has not lived up to these lofty expectations. While the overall SaaS industry has increased 10x over the last 15 years, it has largely remained unprofitable. A recent study by Thoma Bravo revealed that the average-scaled SaaS company has a 71% average Gross Profit Margin but deeply negative Net Margins (EBITDA).
The main culprit?
Sales and marketing expenses. Every dollar of revenue requires 75 cents of marketing spend. So even before you provision for the cost of research and development and other corporate expenses, you are already in the red.
– Sumanth Raghavendra
Fair enough. Sales and marketing are the holy grail of getting people to adapt to your me-too software. And the real problem is probably not that sales and marketing costs are too high. That is the effect, the cause is probably that the products are hardly differentiated and seldom of outstanding quality.
Another reason Raghavendra states also makes sense:
The other reason for the demise of the original SaaS promise was the glut of capital that entered the market in the ZIRP (Zero interest rates period) years.
– Sumanth Raghavendra
This seems true. Excess of VC capital flowed into SaaS that was then put into too much sales and marketing leading to a glut of marketing SaaS companies, creating a kind of a Ponzi scheme.
But then the post gets murkier and murkier.
According to him, the new age “AI” companies will not be capital intensive but can be run by small teams with little capital. Mmm, hasn’t he heard of Sam Altman’s need to raise 7 trillion in capital to make new chips?
Let’s assume that AI will kill all software companies. But why? Because presumably AI will write better software than us? Well, there is no evidence of this so far. The generative AI on display is very early stage as of now. More and more people are calling out the poor quality of code these AI engines are throwing out, and the fake demo videos going around.
Even if this is true, that AI will be writing software, then let us accept that the team that has the best AI generator will win and the winner will take it all (There is no silver medal). Not only TCS and Infosys, but Microsoft and Google too will be filled up with very few engineers and thousands of robots writing code. He goes on to say that not only engineers writing code but marketers writing SEO clickbait posts and salespeople doing “desktop sales” will also be replaced by robots.
On business models, Raghavendra states that the “seat-based” SaaS pricing model will be replaced by a usage-based one — this is 100% true, and this will be driven by, what he fails to mention — open source. Open source today is already one of the fastest-growing models of software distribution and the consumer benefits by either deploying the software on-prem at a fraction of the cost or paying a usage fee on any generic public cloud.
So what solution does Raghavendra offer to this so-called death wave of SaaS?
“Gird our loins”? That’s it. So after all that build-up, all we are supposed to somehow do is get ready for the tsunami.
Since the post does not give any suggestions, let us explore this on our own. HOW will generative AI change the industry? Let’s look at all the expenses in our Income statement:
- Software development (usually 15%) — Most products don’t require many engineers to build them, the modern tooling is already very good. Maybe AI and robotic code generators can cut some slack? But this is also questionable. The robotic code is usually of poor quality and much harder to debug than code written by people.
- Cloud hosting (usually 15%) — Assuming we save money by getting robots to write code, the robots are going to guzzle a lot of computing. Whether we buy that from OpenAI or we run our own open-source robots, this cost is going up.
- Marketing and Sales (usually 50–70%) — Most of the money here goes to big tech — Google and Meta to get the lead flow. We also pay for a lot of marketing tech SaaS to stalk our users and send them robotic mails, videos, and messages. This is already a nightmare for humans, what more do you want here?
When I look at this, I really don’t see a way to solve the really hard problems.
- Building good products is hard
- Distributing products is hard
- Being differentiated in the market is hard.
AI is not probably going to solve any of these. Then what is AI solving for in the context of software products? AI is good at solving specific problems like translations or image classification — none of these above problems seems specific enough.
Maybe Raghavendra is looking at the demand side, like a new class of products that will be created to solve our workflow problems. What will this robotic text generator do? Write documents, legal contracts, marketing copy? write code? make social cards? pay our bills? If so, are any of the SaaS companies today experts in these fields?
A genuine question to all software / SaaS companies reading this. What do you really need? Better quality software? Better quality leads? Is AI going to automagically solve any of that?
I have been a silent supporter of SaaSBoomi over the years, because I think it is the only genuine community for founders. While I have connected with several interesting folks, I have always been worried about the excessive focus on sales and marketing in the community. Somehow our USP is being able to do cheaper sales and SEO from India (sales arbitrage). It has never been around differentiated products, technology, or quality. I have written more on this in 2020.
Now that SaaS seems to be losing steam because of too many products and dwindling funding sources, I am worried that the message that is given to the founders is to go find the next wave rather than introspect on what will make us build differentiated products. It almost felt that Sumanth had outsourced his thinking to Generative AI, creating a lazy and superficial thesis, complete with a meaningless AI-generated image that somehow is reflective of our times.
Yes, SaaS is probably in trouble and what we need is a “real” conversation. The Indian software product community deserves better.
Editorial Note:
This article was originally published by Rishab Mehta, in response to the article ‘The AI apocalypse is coming.’
A cohort of more than 100 AI-first Indian startups is convening in Bangalore on June 7th for a 3-day workshop that will prepare us better for the future. If you would like to be part of this program, read more here.
