SA国际传媒

Over the years, has put copious effort into branding itself around all things cloud. The firm has a -traded 聽 and a large portfolio of prominent cloud startups.

So, when Bessemer calls its latest annual report 鈥渁 eulogy for the legacy cloud,鈥� it warrants attention.

In its latest report, published Thursday, the firm makes the case that startups are moving out of what it describes as legacy SaaS business models and embracing a new paradigm: vertical AI. While this isn鈥檛 a death knell for the enterprise software companies that have been its bread-and-butter for years, it does mean they鈥檒l have to evolve with the times.

鈥淚t is a big deal, particularly coming from Bessemer. I think it is reflective of just how big of a movement and paradigm shift this is,鈥� said firm partner , referring to the rise of large language models and AI-enabled tools that are reshaping the enterprise software landscape. 鈥淭his is an extinction event for companies that are not on the early side of adoption.鈥�

All about vertical AI

If cloud was the perennial buzzword of years鈥� past, this time Bessemer is pushing to popularize the concept of vertical AI as the next big thing. It defines this as applications that 鈥渢arget the high cost repetitive language-based tasks that dominate numerous verticals and large sectors of the economy.鈥�

Bessemer, which cites over a dozen of its own portfolio companies as vertical AI exemplars, isn鈥檛 the only one highlighting the trend. Late last year, for instance 1聽published a , and listed in the space for healthcare, finance and professional services.

Per Dholakia, some of the most promising startups in vertical AI use cases are well beyond seed stage by now. Many secured initial funding well before the late 2022 launch of established the power of large language models for the masses.

While we鈥檙e probably at least a couple years away from seeing vertical AI IPOs at scale, Dholakia is already seeing early entrants shake up prevailing business models in the enterprise software space. One shift that AI is enabling, he said, is the propensity to charge customers based on work rather than number of users engaging the software.

鈥淵ou鈥檙e seeing people price the delivery of their capability differently,鈥� he said. 鈥淭hey鈥檙e solving a piece of work, and charging per work product.鈥�

Dholakia cites legal tech AI portfolio company as an example, with the firm currently able to charge personal injury lawyers for demand letters created with its offering.

Exits and valuations

While Bessemer is enthused about the promise of vertical AI offerings, the firm is also cognizant of the reality that startups in the space are mostly a ways from exit. Meanwhile, there remains a vast swathe of still-private, later-stage enterprise software companies formed before the all-about-AI era.

The backlog of not-yet-exited unicorns, in particular, is amassing amid a sluggish period for IPOs and large M&A deals involving private, venture-backed software companies. And with the traditionally languid summer IPO season upon us, the slow streak looks set to continue.

Dholakia, for his part, said he鈥檚 optimistic the offering pace will pick up, likely by next year. For the past couple years, he said, the software startups have held off going public, given the post-2021 market contraction, several quarters of subsequent belt-tightening, and pressure to update with AI capabilities.

As a result, there鈥檚 now a 鈥渙nce in a generation, maybe once in a lifetime backlog of companies鈥� that are worthy of going public but are still private, he said.

Hopefully, soon we鈥檒l see them making their way to market.

Related reading:

• The 49-Year Unicorn Backlog

Illustration: