The bygone era of Killer Apps

The Internet-era world was so simple! For a software company, the most that was needed for years- and at times, decades-long dominance was to build a great product with a great idea before anyone else. 

The table above captures only a slice of the story, and folding in global giants like WeChat, Shopify, PayTM, Grab, Kareem, Spotify, MercadoLibre, Zillow, and Revolut makes the case for first-mover advantage even more striking. Back then, a brilliant idea paired with good execution was the golden ticket—think of how these names carved out their domains and held the lead for years. In an era when ideation ruled, spotting a great feature or product’s first arrival in a space was enough to unlock massive gains, with few extra bells and whistles required. History fans love to point at these wins, assuming today’s software champs can just ride the same wave with useful, new, AI-powered tricks up their sleeves.

But, the history buffs drooling over software’s past, mesmerized by and frequently quoting the feature/idea-based success record, need a reality check—this isn’t some centuries-old saga with predictable reruns. The software and services sector is a toddler, barely three decades old, and the pattern-spotters clutching their tales of recurrence must analyze the drivers that caused winners to enjoy certain levels of market shares, pricing power, and margins. One only needs to compare the table above with the one below to see how much has changed.

In some key features like “Deep Research” or Reasoning Models, so many began so together that it is impossible to assign a pioneer.

Not Innovating is Not an Option, But Innovation Itself Won’t Be Salvation

Innovation remains critical for survival in the tech world. Companies like Hynix, TSMC, ASML, and Nvidia prove this. Hynix’s High Bandwidth Memory (HBM), TSMC’s CoWoS packaging, ASML’s EUV lithography, and Nvidia’s GB200 are technological marvels. These innovations are so advanced that rivals will struggle to catch up for years if not decades. In such cases, it is possible for investors and other stakeholders to focus simply on product features and announcements while considering their relationships with the innovators. 

Unless one believes in software companies wowing those pay them simply because of the way things happened before (“historic patterns”), these companies' ability to seek premiums for innovations or new AI features has to be different compared to before. A personal example to make a point.

As an amateur photographer, I’ve been smitten by Adobe for decades. I’ve paid for its software for over 20 years, and there were years where I likely spent more time on Adobe’s tools than on Microsoft Excel—surely, I paid them more, too! My expertise with their suite, especially Lightroom, is a habit and a convenience, honed through years of editing and experimenting. Yet, despite this deep familiarity, I haven't used Adobe's GenAI features even once when it comes to generating the images accompanying my articles—over 100 by now. Instead, I’ve relied on a dozen different LLM models to provide some color to the write-ups you read.

Why the switch? Adobe’s AI, while impressive in its own right, feels like an extension of a photo-editing powerhouse rather than a tool designed for pure creation. It excels at refining existing images—adjusting lighting, smoothing textures, or enhancing details—but struggles when I throw complex, imaginative prompts its way, like “take this whole article of 3000 words on Manus and create a symbolic image of its themes with Marathi Manus” (see the image with the previous article).

Adobe will catch on, but I do not see myself using it exclusively or even primarily without thinking about alternatives for more and more of my image workflow. And it has caught my attention that I pay more to them than multiple LLM-makers on a monthly basis, despite using the latter tens of times more. Old habits die hard, I guess.

The primary revenue drivers for software companies, their management, developers, investors, and clients were seen in new ideas and features. Without new ideas and features in the GenAI era, most tech companies will dwindle (Apple: please take note). But, whether it is for an OpenAI, Salesforce, or any B2B/B2C type of company, the introduction of nice AI features is unlikely to lead to new revenue drivers purely on the strength of the features. We are in the “instant copyability” world where pure ideation is being commoditized. 

In other words, innovation is table stakes, not a winning strategy. The likes of Adobe and Salesforce can’t afford to stop innovating, but neither can they rely on innovation alone. There are many other moats possible for each one in the strength of relationships, distribution networks, advantages from access to historic/proprietary data, regulations, security, etc. Many software companies will find new drivers from these moats, but far more will likely struggle.