Five trends driving the adoption of Cloud AI technologies, showing reality outpaces hype

Bessemer Venture Partners is one of Silicon Valley’s most prominent venture firms and one of the earliest to make a bet on SaaS and cloud technologies. Historically, It has parlayed the firm’s expertise into annual trend reports such as the State of the Cloud. In 2023, with the rise of ChatGPT, the company declared that language model-powered AI was here to stay. And it was right. For this year’s State of the Cloud, Bessemer believes Cloud AI has supplanted the legacy cloud and that the reality about the technology is now outpacing the hype.

“We’re at that moment of a tech cycle where everybody’s paying attention,” Kent Bennett, Bessemer partner and lead author of the 2024 State of the Cloud report, tells VentureBeat. “But the actual data that we’re seeing on the ground, on the reality of the speed at which this technology is being adopted by our existing portfolio companies, used to form net new companies—new companies that actually have customers and revenue and growth profiles and efficiency, that are truly unlike anything we’ve ever seen.”

All full-time investors with the VC firm participated in the research, providing an analysis and prediction on what’s happening with the cloud over the next few years. Bennett admits the 62 global investors who participated aren’t statistically significant but says this focus group provides deeply focused insights on one topic versus if there were 100,000 investors and went “super wide around a bunch of topics.”

Nevertheless, he and his co-authors believe the takeaway for fellow investors, founders, CEOs and executive teams is that “if you’re building cloud software right now and you’re not thinking about how to leverage [AI], you’re likely missing it.” They cite how companies over the past two years have become AI-native firms armed with full knowledge of language model technology and how they can use it to improve their products. “If you are a cloud software company and you’re not making an attempt to incorporate this technology into your product, you’ve actually got a real problem because somebody out there is, and they’re going to come for you.”

Here’s a look at the five trends outlined in the State of the Cloud report Bessemer believes are shaping the future of the Cloud AI economy.

Big Tech will do battle over AI foundation models

“Foundation models are the new ‘oil’ that will fuel downstream AI applications and tooling,” the report authors write. They state that these models will be weaponized by Big Tech companies to secure market dominance. In 2023, 60 percent of total AI dollars was spent on foundational model companies, with OpenAI, Anthropic, Mistral, Cohere and others raising $23 billion in investment and a collective market cap of $124 billion. It’s certainly no paltry sum. And with Microsoft, Google, Amazon, Nvidia and Oracle taking significant stakes in these model providers, it’s likely that we could soon see what Bessemer describes as the “battle-of-the-century” brewing.

Bennett doesn’t think Big Tech companies will form an oligopoly around foundation models. Instead, the average opinion at Bessemer posits that it will be a hybrid scenario. “We think the bulk of the value captured in the model layer will be captured by a combination of the big incumbents—like the current cloud vendors—plus a few of the household name layer companies,” he clarifies.

The firm acknowledges this may not be a worldwide event, especially with some governments pursuing sovereign AI. “One nuance on that is that there may be many geographic winners as different regions and countries believe they need to own their own model layer, and they’re not going to outsource artificial intelligence to the United States necessarily. So there may be more flourishing as a result of that. But for the most part…the margins there will be owned by the big companies.”

Bessemer predicts the battle for AI models will be a “critical ‘land grab’ that determines which Big Tech companies reign supreme within the cloud and compute markets in the coming years.”

AI has turned us into developers

While the AI boom seems to require developers to learn new languages, frameworks, infrastructures and tooling, we now know that technology has made programming easier thanks to tools like GitHub Copilot, coding assistants from Amazon, Oracle, Mozilla and Ubsoft, and AI engineers from Cognition, opening the field up to more people.

Bessemer predicts that by 2030, every human with a computer and a phone will have “significant developer capability,” resulting in a dramatically reduced average age of technology startup founders. In addition, the firm believes most corporate software developers will become the equivalent of software reviewers. Development costs will fall, and the more capable experienced developers are, the higher their salaries will become.

What’s causing this evolution of the AI developer? The growth and innovation of code copilots, for one. Secondly, the so-called “graduation motion” of copilots featuring agentic search and generation functionality will have outsized value. And finally, code-language reasoning will remain at the core of AI activity.

Multimodal models and AI agents will change how we work with software

Static text-based chatbots are a thing of the past. Bessemer states that the future rests with multimodal models. Writing has been replaced by vision, hearing and speech in AI applications. The firm is placing a specific bet on voice AI and expects these types of apps to have “breakout growth” over the next 12 months.

Moreover, people will see intelligent and autonomous AI agents handling tasks and managing workloads. We already see these agents used in areas like customer service, project management, etc. Expect stronger reasoning to make these bots handle more complex workflows. Also, we might soon see novel approaches to improve agents’ handling of chain-of-thought reasoning, self-reflection, tool use, planning, and multi-agent collaboration.

Vertical AI could dwarf legacy SaaS with new apps and business models

The penultimate trend looks at how a focus on verticalized AI might open up new business models and application ideas than what was possible with legacy SaaS. “Our strong consensus is—and this might be one of the points where we had the most consensus—is that vertical AI is a bigger [enterprise value ] opportunity than the legacy vertical SaaS,” Bennett explains. He claims vertical AI technology is selling something where “the value proposition is comparative to very expensive labor,” and although it’s not supplanting labor, it’s delivering something that “is valued on the order of human labor.” It’s not going after software budgets but rather the Return on Investment that the revenue generates or the significant cost reductions.

“It’s sort of like every business in the world can now afford to have some level of massive extension of their labor productivity for a set of highly repetitive tasks that can be executed perfectly,” Bennett says.

He elaborates further by saying that the way AI companies are pricing their products compared to legacy SaaS is on par with each other but happening at a much faster rate than expected.

“What we’ve seen in the data from the early startups we’ve gotten involved with where we have access to what’s actually happening on the ground, we’re seeing it play out exactly like that: The pricing of these products relative to their legacy sort of peers in the same verticals, is already at the same level. In some cases, [it’s] bigger, which is just wild because it’s been two years. This pricing is far from mature,” Bennett states. “In many cases, these products are not nearly as penetrated as their legacy peers. And so, we don’t want to overstate it without actual data to reflect this, but we feel like, in many cases, you don’t have to take a radical leap to say that the size of these opportunities could be 10x their peer SaaS counterparts in many of these verticals.”

AI might resurrect the consumer cloud

It’s been eight years since the consumer cloud space saw a major exit (acquisition or IPO). Bessemer’s partners think AI will change this pattern. The firm defines consumer cloud as “companies that provide cloud-based storage, compute and digital applications directly to individual consumers, including, at times, concurrent B2B and ‘prosumer’ offerings.”

The report points out how LLMs’ multimodal capabilities are creating disruptive opportunities in every category of consumer cloud. For example, OpenAI’s ChatGPT is drawing as many monthly visitors to its site compared to Reddit and X, and it’s not the only one as Anthropic’s Claude and Google’s Gemini are also gaining steam.

Quite a few consumer-focused AI companies are raising large sums, too, at high valuations. Startups such as Perplexity, Character.ai, Midjourney, Suno, and Luma come to mind. Bessemer points out these firms demonstrate how LLM-native apps can attract and retain large numbers of dedicated users and threaten to displace modern incumbents.

The investment group predicts “multiple consumer cloud IPOs” will occur within the next five years.

Business leaders: AI isn’t hype anymore

Some of these trends listed aren’t surprising, as we’ve already seen them happening this year. Bennett tends to agree but reminds us that it has seen a lot for a firm with a storied past that has been around for the dotcom, mobile, and SaaS waves. But, all that “pale in comparison to what we’re seeing so far” when measured just by the number of actual real companies with real revenue and real customers. Bessemer’s front-row view lets it bear witness and understand that, unlike past waves where hype eventually outpaces reality, this time, it’s different. “The reality is running faster than the hype…and the hype is crazy.”

He shares that business executives reading this report should walk away knowing that startups in this fledgling industry are already making their moves to engage the enterprise. However, organizations should exercise patience because it will take some time before these AI solutions have reached their full potential. Bennett urges companies to design for the future “in very modular ways” meaning that when baking AI into internal tools, build it so that when a superior offering comes out, the technology can easily be inserted, saving time and resources.

It’s also advised that organizations maintain a watchful eye on the AI space, researching and trialing new startups that might have convenient and suitable solutions as early as what’s appropriate.

Along with Bennett, Mike Droesch, Sameer Dholakia, Talia Goldberg, Janelle Teng, Caty Rea, Lindsey Li, Maha Malik, Bhavik Nagda, Aia Sarycheva, and Alex Yudtiski co-authored Bessemer’s 2024 State of the Cloud report.