Will AI create massive opportunities for startups?

Conventional wisdom says so. But, maybe not.

🍾 Happy new year! It’s Ravi, creator of popular frameworks like the Product Competency Toolkit, the Product Strategy Stack, and Narrative, Commitments, and Tasks. I have a halfway decent track-record of making and keeping new year’s resolutions. This year, I’ve resolved to share more often about my experiences building products & teams.

As part of that, I’ve moved my newsletter over to Substack. The best part of having a newsletter has been the conversations I’ve had and the people I’ve met. With all the social features on Substack, I hope to have more of those conversations. You can also reach me on LinkedIn and via email.

---

By now, it’s clear. We’re at the beginning of a major platform shift. The last big platform shifts (the Internet and mobile) created fertile ground for companies like Google, Amazon, Spotify, and Uber to capitalize on a moment of disruptive change.

That's why VCs love a good platform shift. In fact, they love platform shifts so much sometimes they seem them in their dreams. Crypto, anyone? 💸

So, $50 billion dollars went into AI last year—anticipating the wave of successful startups that will disrupt incumbents.

But, big companies seem to be reaping the benefits of AI, while many startups are struggling to turn hype-induced user acquisition into retained customers.

Why?

Maybe because:

1. Big companies learned from the past and were better prepared for this shift. (I'm not so sure: Google invented the "T" in GPT [transformers] , but ceded the moment to OpenAI.)

2. AI models are so expensive to build & train that startups can't do it. (I don't buy it: many AI startups have raised enough capital to train the most intensive LLMs.)

3. Big companies have the large, proprietary data sets needed to train good models. (For sure, but that didn’t stop OpenAI, Anthropic, Inflection, and others. There is enough public, crawlable data to create remarkable models.)

4. Big tech companies have gotten too big to disrupt. (That's definitely a challenge, yet an entirely new product, ChatGPT, was the fastest growing app of all time.)

5. Customer acquisition has gotten too expensive thanks to Google, Facebook, and Apple. (Also a challenge, but the AI hype cycle has driven an unprecedented amount of word of mouth traffic to new AI products.)

Technological shifts vs. distribution shifts

Casey Winters

has a really good article, "On Platform Shifts and AI" where he draws an important distinction between technological platform shifts and distribution platform shifts:

What separates a major platform shift from a minor platform shift is a platform shift that enables both a technological shift (new ways of making things possible) paired with a distribution shift (new ways of reaching people with it).

Google is the prototypical example. They benefitted from the technological shift of massive of amounts of information becoming crawlable via the Internet, and they drove a distribution shift by changing how people discover and access that information.

So far, AI has led to a technological shift, but not a distribution shift. That may change with the GPT Store or other means of leveraging LLMs for distribution (such as organic or paid citations included in generated results).

This will be an important long-term factor, but I'd argue that AI hype / word of mouth has created a significant (albeit temporary) distribution opportunity.

So what else could it be?

The arithmetic of disruption

It may come down to simple math: addition and subtraction.

In the 1990's, Harvard professor Clayton Christensen coined the term "disruptive innovation":

Disruptive innovations are NOT breakthrough technologies that make good products better; rather they are innovations that make products and services more accessible and affordable, thereby making them available to a larger population.

Christensen goes on to explain, “disruptive innovation describes a process by which a product or service initially takes root in simple applications at the bottom of a market—typically by being less expensive and more accessible—and then relentlessly moves upmarket, eventually displacing established competitors.”

Disruption starts with subtraction.

The Internet and mobile platform shifts were defined by subtraction.

For example, the Internet required companies with large assets, such as brick-and-mortar facilities, to rethink how their businesses create & capture value. Pre-Internet companies needed to subtract the very thing that made them successful—their expensive, hard-to-replicate physical footprint.

This was even true of pre-Internet tech companies. Companies, like Adobe, had built their business on powerful, "single player" Windows and Mac applications—they needed to rethink how to deliver value in a “multiplayer” web browser.

At the time, Adobe Photoshop was one of the largest codebases in the world. How could they subtract all of that power and whittle down Photoshop to work in a browser? They were slow to adapt, and left the window open for Figma.

The mobile platform shift was also subtractive. Companies needed to simplify and refine their desktop products into portable experiences. Despite an early start, Microsoft couldn’t make the necessary deletions from Windows Mobile, and they ceded the opportunity to Apple and Google.

Subtracting is really hard. Blackberry knew that touch screens were the future of mobile for years, but couldn’t bring themselves to delete the chiclet keyboard that had once been their defining feature.

Thinking, fast and slow. Like way too slow.

Why is it so hard? I think it has to do with loss aversion, the psychological phenomenon where people tend to prefer avoiding losses over acquiring equivalent gains. The pain of losing something we once had is more acute than the pleasure of getting something we want.

Companies are simply communities of people that stand to gain or lose individually and collectively. People can’t turn away from the golden goose even when it’s clear that bird’s day is done.

That’s why stories of reinvention, like Netflix’s shift from DVD rental to streaming, are so powerful. Vision isn’t the hard part. Avoiding disruption requires an extraordinary and rare level of fortitude.

Many big companies suffered during the Internet and mobile platform shifts. They could not muster the fortitude to make the decisions they knew they needed to make. This created the opportunity for startups to flourish.

As Christensen put it, prior platform shifts created the perfect climate for startups to “take root in simple applications at the bottom of a market” and then “relentlessly move upmarket, eventually displacing established competitors.”

The new math

Will this disruption happen again with AI? The answer isn’t so clear.

AI is an accelerating technology. It makes established products even better. We can already see how this is playing out.

Take Adobe for example. While Adobe was not able to make the difficult deletions necessary to beat Figma to the cloud collaboration market, they have moved quickly to layer generative AI into their products. This includes adding AI features into existing products, like Generative Fill and Generative Expand, that integrate image generation models into Adobe Photoshop’s existing toolset, and Adobe has launched new standalone products like Adobe Firefly. Adobe is still far from parity with Figma’s collaboration features, but they have quickly leapfrogged on AI features.

Established companies were at a disadvantage during the Internet and mobile platform shifts, because those shifts threatened their core. Today, AI is helping established companies reinforce their core.

Microsoft is the best example. They are reinforcing their core productivity & business offerings by layering AI into Microsoft Office, Edge, Bing, and Teams. After years of losing developer mindshare, Microsoft built a strong position in developer platforms by acquiring Github and launching VSCode. Today, that position is getting reinforced with Github Copilot.

Google has been slower to execute, but that may change in 2024. By the end of the year, I believe Google’s search franchise will be stronger than ever with the addition of generative search results and Google Adwords placements into those results.

So, what’s a starry eyed startup to do?

The path for AI startups won’t be easy. Not only is AI accretive to established products, big companies have the other advantages we touched on at the beginning: massive, proprietary data sets, resources to build & train the most advanced models, and cheap distribution at scale.

We’ve already seen casualties inflicted by the fast-pace of OpenAI product launches. In the face of these challenges, AI startups must take a judo-like approach to strategy:

In the martial art of judo, a combatant uses the weight and strength of his opponent to his own advantage rather than opposing blow directly to blow. Similarly, smart start-ups aim to turn their opponents’ resources, strength, and size against them.

New AI startups face important strategic questions:

• What are the riskiest opportunities? Unless they have a clear advantage, startups should avoid competing directly with AI platforms, like OpenAI, or hard-to-replicate products, like Microsoft Office or VSCode.

• Conversely, what opportunities are big companies ill-equipped or uninterested in pursuing? For example, Harvey is building generative AI models for law firms and other professional services companies—a meaningful opportunity with limited competition.

Next week, I’ll share a set of observations & principles AI startups can use to win in an intensely competitive environment.

What do you think? Do startups face a bigger challenge competing in the AI platform shift than the Internet and mobile platform shifts? How can AI startups best compete?