With a blog titled "Head in the Cloud(s)," you'd think I'd be all in on cloud computing, distributed computing, large data centers, etc. Yes, the shift to cloud computing has been a net positive for specific applications. It is easier to get up and running, more scalable than on-premise computing, and financially more accessible to track.
That said, the emergence of large language models (LLMs) and generative AI creates a once-in-a-generation opportunity for certain on-premise, non-cloud applications.
For all the buzz around ChatGPT, Perplexity, etc., there's a quieter movement to deliver on-premises AI, leveraging several open-source packages in current use. Without naming specific vendors or packages, I submit that the (relatively) quiet development of on-premise AI is both a significant investment opportunity and a means by which organizations with security concerns or just large volumes of data can take more substantial advantage of the AI revolution.
Let's use the insurance industry as an example. Several years ago, incumbents were sprinting to develop glossy new interfaces for customer acquisition, as well as a counter to the emergence of the new "insurtechs" that threatened to "digitize insurance." Years later, it appears that much of that "insurtech wave" was little more than putting a digital front end on existing processes and targeted more at "edge" applications such as customer acquisition vs. "core" applications in insurance, most notably underwriting. The few insurtechs that managed to go public (Lemonade, Root) have seen their stocks largely crater in the public markets since IPO (LMND: -77%, Root: -88%), as the level of real disruption they posed to the industry was minimal, and thus their profitability.
Fast-forward to today, and with on-premise AI, an opportunity is emerging to tackle one of the core value propositions that incumbent insurers still have a handle on: underwriting. Decades, if not centuries' worth of data exist almost entirely on-premises, which can be used to train an enterprise AI to make better underwriting decisions. Because of the media where that data resides and the sensitivity, that information will certainly NEVER be shared to the cloud.
THAT is the opportunity.
Stay tuned.