We’re joined by Vijay Raina, a leading expert in enterprise SaaS technology, to unpack the seismic shifts AI is causing in the software industry. Drawing on insights from tech leaders at the forefront of AI development, we’ll explore how the very economics of software are being rewritten. Our conversation will cover the evolving role of the software engineer in an age of AI-generated code, the potential rise of cheap, disposable applications, and the profound career displacement that may follow. We will also delve into the conflicting futures for SaaS platforms envisioned by top CEOs and what this disruption means for investors and founders trying to navigate this new landscape.
At companies like Anthropic, some engineering leads report they now mostly edit AI-generated code. How does this fundamentally change the day-to-day role of a software engineer, and what new skills should they prioritize to remain valuable? Please elaborate on the productivity metrics you’ve seen firsthand.
It’s a complete paradigm shift, moving the engineer from a creator to a curator or an editor. The mental energy is no longer spent on syntax or boilerplate code; it’s focused on high-level architecture, logic, and refinement. I’ve seen this firsthand in the stories coming out of Anthropic. Their team built a new tool called Cowork in just a week and a half, a task that would have taken months before. This wasn’t a case of AI helping a little; it was almost entirely built with their Opus model. So, while engineers are still needed for that final 10% of the work—the critical oversight and editing—their value is now in their ability to prompt, guide, and validate the AI’s output. The new essential skills are strategic thinking, sophisticated prompt engineering, and an almost architectural ability to see the entire system and direct the AI to build it piece by piece.
A future where custom software is cheap and disposable—like an app built for a single meeting—has been proposed. What are the first practical signs we’ll see of this shift, and how should existing SaaS companies pivot their business models to survive this potential economic disruption?
The first signs are already appearing inside tech-forward companies as internal tools. Instead of a multi-week project to build a simple dashboard or a data-entry form, teams will start generating them on the fly for specific tasks and then discarding them. The idea of building an app for a single meeting that costs just a few cents to create is the ultimate expression of this. For existing SaaS companies, this is an existential threat. The business model of amortizing development costs across millions of users simply collapses if the cost to build becomes negligible. To survive, they must pivot from selling a static product to selling access to the very systems that create these disposable solutions. Their value will lie in their proprietary data and the specialized, fine-tuned models they can offer, which can generate more accurate and relevant custom software than a general-purpose AI could.
Given the prediction that AI could soon handle end-to-end software development, what specific steps should tech leaders and educators take now to prepare the workforce for the massive career displacement you foresee? Please share your thoughts on the societal adjustments that will be necessary.
The most alarming part of this is the speed. Some are suggesting we could be just 6 to 12 months away from AI handling the entire software development lifecycle. This is not a distant future; it’s happening now, and as Dario Amodei warned, the awareness of its magnitude is dangerously low. Tech leaders must immediately begin aggressive internal reskilling programs, moving their engineering talent “up the stack” from coding to systems design, AI orchestration, and product strategy. Educators need to radically overhaul computer science curricula, shifting the focus from learning specific programming languages to mastering creative problem-solving, logical reasoning, and how to effectively collaborate with AI systems. Societally, we need a serious conversation about what happens when entire, well-paid careers are rendered obsolete. This disruption is so fundamental that it forces us to re-evaluate the nature of work itself.
Satya Nadella has suggested SaaS applications will “collapse,” while Jensen Huang argues their data will make them more valuable. Where do you see the truth lying between these two views? Explain which types of SaaS platforms are most at risk and which might thrive by leveraging their data.
Both are correct; they’re just talking about different ends of the SaaS spectrum. Nadella’s prediction of a “collapse” is aimed at the vast number of SaaS tools that are essentially just user interfaces for common tasks—think basic project management apps, schedulers, or generic CRMs. AI agents will simply absorb those functions, making the standalone app redundant. Huang’s vision, however, applies to platforms that are deeply embedded in specific industries and sit on what he calls “data goldmines.” These companies have unique, proprietary data that is incredibly valuable for training specialized AI. Their software isn’t just a tool; it’s the ecosystem that generates and refines that data. These platforms will not only survive but thrive, as their data becomes the most critical asset for building powerful, industry-specific AI.
The Morgan Stanley SaaS index has shown steep declines, suggesting investors are already wary. For venture capitalists and founders, how does this AI-driven shift change the calculus for building and funding a tech company today? What does a defensible business model look like in this new paradigm?
The market is clearly spooked. The divergence between the struggling SaaS index and the broader Nasdaq shows that investors are no longer confident in the old model of building a piece of software and scaling it. The calculus has completely changed. Previously, a strong engineering team and a clever feature set could be a defensible moat. Now, the code itself is becoming a commodity. For founders and VCs, the key question is no longer “What can your software do?” but “What unique data do you have access to?” A defensible business model today is built on a proprietary data flywheel, a strong community that generates that data, or a unique, complex workflow that is difficult for a general AI to replicate. The moat has shifted from the code to the context and the data that surrounds it.
What is your forecast for the software development industry over the next two years?
Over the next two years, I forecast a period of intense and often chaotic transformation. We’ll witness a great bifurcation in the SaaS market. Many undifferentiated, feature-based SaaS companies will face immense pressure, leading to a wave of consolidations and failures as their functions are absorbed by large AI platforms. Simultaneously, companies with unique data assets will see their valuations soar as they become the new gatekeepers of value. For the workforce, it will be a period of high anxiety and rapid adaptation. The demand for traditional coders will plummet, while a new class of “AI orchestrators” and systems thinkers will become the most sought-after talent in tech. It will be a turbulent, disruptive, but ultimately formative period that reshapes the entire industry.
