Subscribe | Advertise

Bubbles in the AI Startup Market and the Curious Case of AI Scribes 

As has been reported ad nauseam by tech pundits and LinkedIn talking heads alike—the AI market has experienced explosive growth over recent years. The sheer quantity of AI solutions coming to market has skyrocketed in the past year alone, with ~70,000 AI startups in the market today. In dollars and cents, recent reports show that global AI investments increased by over 40% between 2022 and 2023, with AI-related startups raising more than $19 billion in Q3 2023 (28% of all venture funding). That trend has continued into 2024 with Pitchbook reporting $10.5 billion pouring into AI startups in Q3 2024 (incl. OpenAI’s $6.6 billion raise) and Elon reportedly in talks to raise billions in additional funding for xAI.

This frenzy is fueled by technological advances, increasing data availability, and demand across all verticals for faster, more intelligent solutions. AI promises to unlock a range of new capabilities, features, and possibilities that were once unimaginable. For businesses, AI offers tools that automate complex workflows, enhance decision-making, and create innovative customer experiences. It touches nearly every industry—healthcare, retail, transportation, and more—making it a focal point for companies looking to stay competitive in a rapidly evolving digital world.

However, all that excitement comes with its own set of challenges. As AI becomes more accessible and capable, competition among startups heats up. New entrants coming to market today find it increasingly difficult to stand out among their peers—not only to their investors but also to their customers. So, how should startups navigate this challenge?  Maybe we can look to healthcare—of all places—for some ideas.

The Curious Case of AI Scribes: A Snapshot of AI in Healthcare

While some claim that healthcare is tech-averse and reluctant to change (and we can admit that the sector hasn’t historically been a ‘trailblazer’ when it comes to tech adoption), AI has been positioned by many as the spark of efficiency needed in a desperately broken system. From physician shortages to expensive, complex, and opaque processes like drug reimbursements and healthcare coverage—there’s plenty to be fixed and tinkered with in the healthcare space. Although AI is often pitched as a salve for medicine’s most serious and complex issues, the relatively unassuming task of transcribing and documenting clinical encounters appears to be the first to see substantive change. Here come the AI scribes. 

AI scribes look to tackle the systemic issues of clinician understaffing and burnout by automating the documentation of patient interactions and much of the administrative work that follows each encounter. While these tools promise to ease the administrative burden on healthcare providers, the space is already crowded. Some market maps list over 50 companies offering some form of an AI-powered scribe either as a standalone offering or one component of a broader platform. For some of these startups, capital is easy to come by: Suki recently announced a $70M fundraise and its competitor, Abridge, is rumored to be raising more than $250M at a $2.5B valuation (a nearly ~12.5x jump from their valuation in late 2023). For others, there has been no shortage of struggles. Nearly 30% of AI scribe startups have either been acquired or gone out of business between 2021 and 2024, with a significant number folding in the past two years. For example, Augmedix is being acquired by Commure for $139M, and Robin Healthcare & Sopris Healthcare both closed their doors in 2023 and 2024, respectively. 

For those who have lived to fight another day, we suspect that the challenges aren’t going to get any easier in the coming months. As we’re beginning to see in this sea of AI scribe offerings, differentiation is scarce and switching costs are low. Kaiser Permanente is a great real-world case study. One of the largest health systems in the country switched from using Nabla to Abridge across its organization, and the feedback from providers indicated little perceived difference between the two services. Similarly, at the University of Pittsburgh Medical Center (UPMC), where doctors tested Nuance and Abridge in six-week pilots, the response showed a lack of clear differentiation. Rob Bart, UPMC’s chief medical information officer, noted that "the feedback we got [between the two solutions] was almost indistinguishable."

In this scenario, startups are left with three options: win on price, differentiate through unique features, or get acquired. Many are trying to win the pricing war, but how sustainable is a race to the bottom in developing long-term competitive positioning? Companies like Innovaccer and Commure are offering free versions of their scribe software in a bid to lure clients away from competitors. Others, like Abridge and Ambience, are diversifying their offerings. Abridge has launched an AI tool aimed at helping nurses, while Ambience offers a secondary product that automates billing and coding. Suki is targeting Federally Qualified Health Centers (FQHCs) as a unique entry point to the market. Despite these efforts, for many, consolidation seems inevitable.

Adding further pressure to the space is the fact that legacy electronic health record (EHR) systems like Epic and Elation, and Big Tech players like Microsoft, Google, and Amazon are starting to roll out their own AI-powered healthcare tools. Epic, in particular, has plans to integrate a suite of AI features, including scribes, revenue cycle management tools, and predictive care solutions in the coming year. For smaller startups, this presents an existential threat. As if companies like Abridge and Nabla weren’t big enough challengers, the industry behemoths and deep-pocketed Big Tech players are hot on their trail. 

The Broader AI Landscape: Parallels and Predictions

The challenges faced by AI scribe startups are mirrored across the broader AI landscape. As the first wave of the GenAI hype cycle cools, thousands of AI startups are vying for attention and funding, but deal volume is declining as investors become more selective, concentrating capital into fewer, larger rounds. Big players are quickly swallowing up smaller competitors, seeking to consolidate their market position through acquisitions and have managed to skirt antitrust probes by taking an acqui-hire approach. It’s a race against the clock as many startups approach the end of their financial runways without having secured their place in the market.

According to PitchBook, AI startup deal volume dropped by nearly 25% in 2024 compared to the previous year, despite the continued influx of new companies. At the same time, giants like OpenAI, Google, and xAI are expanding their market share, benefiting from their established brand recognition, superior infrastructure, and deep pockets. As competition heats up, the likelihood of consolidation continues to rise, symptoms that we’re already seeing in the healthcare AI scribe arena.

The Path to Survival: Solving Real Problems

So, who’s going to win this AI arms race? Suggesting that startups simply "differentiate" themselves would be an oversimplification. The key to long-term survival in AI—particularly in healthcare—is solving specific, pressing problems that a company is uniquely equipped to handle. This means identifying pain points that are under-addressed and developing novel solutions uniquely suited to address them. By focusing on solving deep and specific problems that are outside the purview of the slower incumbents, startups can create products that are immediately distinguishable from what’s already out there. With highly performant models more accessible than ever and technology advancing at a breakneck pace, the first question should always be around the strategy for sustaining a long-term competitive advantage.  

In healthcare, that might look like more patient- and provider-centric innovations that solve narrow, acute problems along the care and treatment journey. It might involve solving for the unsexy, smaller-scale, but infinitely more painful problems of today before undertaking the amorphous moonshots of a distant future. And it just might be time to move away from singular AI solutions that can “do it all” and towards solutions that can, for lack of a better word, do something incredibly well.