The proofs of a seismic shift in the innovation landscape are mounting. Historically, the image of the lone inventor or small team working out of a garage symbolized the spirit of technological progress. Companies like Hewlett-Packard, Apple, and Google famously began this way, embodying the ethos of startup culture where passion and creativity flourished despite limited resources. However, the GenAI era is transforming this narrative, with innovation increasingly becoming the domain of well-funded giants rather than scrappy startups.
With Adept (https://www.adept.ai/blog/adept-update) recently, post Stability AI and Inflection AI earlier, we are witnessing well-funded and high-profile start-ups with great teams and ideas undergoing back-breaking upheavals exceptionally early in their business cycles. This trend is not merely about better financial packages, stealth acquisitions, or more resources; it speaks to a deeper shift.
In the pre-GenAI era, well-funded startups had the luxury of multiple funding rounds to refine their ideas and achieve market fit. Despite setbacks, these companies persevered, driven by the founders' passion and the investors' belief.
Not now. Notwithstanding substantial initial funding, a rising number of startups are running out of money quickly, exemplifying the power of the hardware providers. The ability to continue is not just about access to capital but also access to data, compute prioritization, and the need to work on more than one or a small number of things for successful GenAI-era product builds. Ideas or actions that take small efforts to build require a short time to be replicated and enhanced by others in the world of machines doing most of the backward engineering.
More strikingly, founders who once thrived on the freedom and culture of startups are increasingly opting to abandon their ventures to join established tech companies. This shift suggests a realization that GenAI ambitions cannot be fully realized within the traditional startup model.
The scale and nature of GenAI-era innovation investment are akin to building a bridge: effectively continuous and substantial funding before the possibility of returns much later. One cannot construct a skyscraper with each floor funded with a new funding round, and the same concept is reaching the shores of tech-land. Of course, if the eventual returns do not come, the bubble wailers would be right about the wasteful nature of the current investments.
The last point, repeated in our previous posts, is the need to mash data and go across applications to extract “intelligence.” GenAI, almost definitionally, is a result of understanding new interrelationships across fields, which in software terms is by going across the capabilities and data that reside in individual applications at present. LLMs started the trend of mixing everything up, and the Agentic era is making it clearer.
