The Death of Non-AI Startups

In ten years, we will not use the word AI.

I do not mean that the word AI or artificial intelligence or even Generative AI will not exist. We will still use these words much like we use words such as microprocessors, the Internet, and cloud. The concepts are important, but for the general public, there is no reason to talk about it.

Take the cloud for example. When Amazon announced “the cloud” in 2006, it was a niche concept that only mattered to a small group of techies. Now the cloud is everywhere, letting us stream music, games, and movies, host video calls, store our documents, and download apps.

But no one talks about the cloud. The cloud is not the thing. What I mean is the cloud is merely a value delivery mechanism to serve you a song, a call with a friend, your travel reservations, or the latest viral game. No one cares how that information got to your phone or laptop.

In the same way, in the next decade AI will simply be the means to deliver value. AI is not the thing.

Of course, the readers of this newsletter do care because we are building startups. Even if you are not technical, you still care about the technology and definitely rely upon the cloud to build your startup. And not since the start of the cloud has a technology wave hit as hard with as much impact as AI. It is literally all anyone is talking about.

TechCrunch noted that in the first half of 2024, over $35.5 billion was invested into AI startups. Five of the six venture rounds of over $1 billion were raised by AI companies including CoreWeave, Scale AI, and xAI. If we measure by startups tagged as “Artificial Intelligence” in YC cohorts, the Winter 24 batch had 63% of the 240 in the group, and increase from 51% in the Summer 23 batch and 29% before that.

How does that stack up against total VC funding? Global startup funding last quarter totaled $79 billion, of which $24 billion went to AI startups, or 30% of all dollars invested according to CrunchBase. Average deal sizes have also increased, with average deal size of $28.9M, up 55% versus $18.6M in 2023.

There are no reliable sources on the number of actual AI startups globally. However, a review of several market maps created by CB Insights included 90 large language model operations (LLMOps) companies in October 2023 and 335 companies using generative AI to automate content, code & design market in July 2023. This is all within less than a year from when ChatGTP was released.

The interest is not just confined to techies either. ChatGTP reached 100 million users within two months of launching, making it the fastest-growing consumer app ever. Searches for specifically “Generative AI” returned over 500 million results to become one of the most searched terms on Google. Before the launch of ChatGTP, the phrase barely registered. Similar trends can be seen for searches of “AI” with over 15 billion results and currently ranking near the top of all searches.

Despite the pushback from the AI doomers, expanding AI regulation, and growing unease about the reliability of current AI models, AI is not going away. The use of AI is accelerating rapidly in schools (good luck preventing students from using it), in companies looking to automate processes, and in technology use cases where huge efficiency gains have already been demonstrated. Andy Jassy, CEO of Amazon, recently shared that AWS’s AI code assistant tool, Amazon Q Developer, helped migrate over 30,000 Java JDK applications in a few months, saving $260 million and 4,500 developer years.

AI continues to be the hot topic and will continue to do so for the foreseeable future. While the initial buzz from a year ago has faded, AI still finds its way in regular conversations because we are still in the exploration phase of this trend. The pace of new model releases has not relented. More advanced GPUs for AI applications are being released. The output of AI from models and tools has rapidly improved. And we are still at the beginning stretch of where the technology will lead.

At some point though, we will have AI figured out. The use cases will be clear, the past issues with drift and hallucinations will be solved, and the market will declare winners in the AI tools race. Most things we use and encounter will be powered by AI, much like a generation ago when “Intel Inside” powered PC’s.

What does this mean for startups? To be clear, being an “AI startup” is not exclusive to building large language models or AI infrastructure. Most of the AI startups are plugging into AI as a mechanism to deliver more value to customers. This could be a wealth management app using AI for portfolio construction or a fashion app using AI to visualize the fit of an outfit for users. These apps are simply consumers of AI, passing tokens via API’s back to powerful foundation models. Some apps might even use local models, but provided by a vendor or an open-source project to power their apps.

Then there are the startups where AI is helping to accelerate the ability to design, develop, and deploy features to users. These are the coding assistant tools like Copilot and Q Developer as well as up-and-coming tools like Cursor that are helping developers be more productive. In the beginning, it was merely code completion tasks, but now AI is providing full code snippets, modernizing code from older language versions, providing architecture guidance, and automating coding tasks like testing and security reviews.

Lastly, there are startups where AI is reinventing the way work is done. Some have said that AI will be able to act as another “co-founder” or a “chief of staff”, doing work that would otherwise require the focus of one or more people. With the rise of AI agents, entire workflows can be automated such as content creation, lead generation, customer support, supply chain management, etc. This will free up employees to work on higher value and more impactful work as well as reduce the pressure to hire as many people early on when startups are still seeking traction.

Right now, startups are still figuring all of this out. Many of the startups launched in this initial wave of AI mania will die off. A lot of use cases will be discarded, but more interesting ones will arise. We are still exploring what is possible, what matters most, and how to create lasting value. This is when the next generation of iconic companies will emerge, much like the Amazon and PayPal in the rise of the Web, Salesforce and Workday with SaaS, Facebook and LinkedIn during social media. Today is the dawn of the AI startup where value creation, delivery acceleration, and automation create companies that execute faster with fewer resources and better results for customers.

When this moment reaches the tipping point, AI will be a given. We will not need to talk about AI for wealth management or using an AI-driven fashion app because it will just be assumed that somewhere inside the box, AI is making things happen. Think about it this way, do you say you use a cloud-based wealth management platform or that you opened your cloud-based fashion app? The answer is no.

I am not sure when this tipping point will happen. It already feels like we are there in the San Francisco Bay Area which has become the global capital for AI startups. The rest of the world is catching up, but it will take less time than cloud adoption took. Pretty soon the idea of a startup that is not in some way using AI will seem strangely antiquated, like a startup that is still maintaining their own server racks in their home-grown data center.

What if your startup is not using AI in one of the three ways mentioned above? The good thing is the Generative AI has significantly lowered the barrier of entry. Startups do not have to hire data scientists, machine learning engineers, and do lots of manual data labeling in order train models manually. Now you just need to be familiar with API’s and some knowledge of prompt engineering to get started. On the tools side, the apps for content creation, code assistance, and other tasks have gotten much better.

Over the next few newsletters, I will dive deeper into the technologies driving Generative AI and also how some startups are using AI today to drive results. If you are actively using AI today, let me know how AI has made your startup better in some meaningful way.

Mark

Are startup pitch events a waste of time? A LinkedIn post recently caught my attention stating that for startups in Asia, "industry events and pitch competitions are a complete waste of time."

I do somewhat agree. Many pitch events that I have attended are poorly conceived and offer little value to startup founders. The biggest downside is the opportunity cost of pitching, exhibiting, and attending events, when the time could be spent on product and customers.

One consideration though is regarding the value for emerging ecosystems in hosting these events. Few places have the density of a Silicon Valley / Bay Area for startup founders. Even when I got started in NYC, there were few events or opportunities to connect with other founders and investors.

Events help bring the community together. Over time, NYC started to feel more like a real startup hub as more folks started launching startups in NYC. This led to more events, which helped raise the visibility of NYC-based startups and NYC as a place anyone could legitimately build and scale a startup.

Startup events are not just valuable for the ecosystem however. For startup founders, the networking allows them to get connected to resources, peers, and investment. Higher profile events in Southeast Asia like Techsauce and Tech in Asia Summits have done a decent job drawing decision makers from investors to corporate executives to come out and meet with startups.

How should you allocate your time though? Understand where you are realistically in your startup journey. Early on you should focus on building. When you are ready to jump into fund raising mode, it makes sense to spend more time networking.

But regardless of where you are, always be frugal with your time and only prioritize events that have a high probability of providing a return on your time (and money) as I shared in this post.

It is still August, but already in AWS circles we are already talking about re:Invent, our global conference in Las Vegas at the beginning of December. However, a conference that brings in over 60,000 people from around the globe requires some advanced planning.

If you are a startup founder and getting more involved in the AWS ecosystem is an important success factor for you (e.g. you want to partner with AWS or you are building a cloud service that helps AWS customers), this is the conference you must attend. While the full conference pass is $2,099, startups can purchase a pass for $899. I cannot share the discount code openly, but if you reach out to me by replying to this newsletter or sending me a message over LinkedIn, I can provide the code.

After some holiday travels and some time back at home, I am out on the road again in September! First stop is a day trip to Chicago on Friday, September 6^th . If you are free that afternoon, let’s catch up. I also am working on some future itineraries that I will share in the next newsletter. Look forward to seeing you somewhere around the world!