By Lina Irawan 
Initially, the SaaSpocalypse thesis posited an outright "death" or at least a massive, existential disruption for incumbent software providers, a prevailing bearish sentiment in the early part of the year. However, as the AI landscape evolved and market understanding deepened, this extreme bearish view has somewhat morphed. While the notion of complete obsolescence has receded for many, it has been replaced by a less drastic, yet still potentially incorrect, theme: that software companies will face significantly higher customer acquisition costs and diminished pricing power, inevitably compressing their margins and hindering long-term profitability.
Just as Francis Ford Coppola’s classic 1979 film Apocalypse Now immersed audiences in a fictional delirium, so too, perhaps, is the current “SaaSpocalypse now” narrative largely based on a misreading of the unfolding technological revolution. There is, unequivocally, an undeniable competitive threat posed by autonomous AI agents to many high-flying technology winners. The list of companies genuinely vulnerable to AI-driven disruption is indeed extensive and growing. However, in their panicked stampede to the exits, investors have indiscriminately punished numerous software firms, including several that are, in fact, poised to become significant beneficiaries of the AI era. These companies, far from being casualties, are strategically positioned to leverage AI for greater profitability and enhanced pricing power. Three compelling examples illustrate this point: Salesforce, Booking Holdings, and IBM. These giants demonstrate how, contrary to prevailing short-term market fears, certain established software companies are well-equipped to emerge as long-term AI winners.
Salesforce: The Unassailable Data Fortress
The misleading bearish AI scenario for companies like Salesforce held an appealing, albeit superficial, simplicity for anxious analysts. Salesforce, the undisputed leader in customer relationship management (CRM) systems, was wrongly predicted to face obsolescence at the hands of large language model (LLM) companies such as OpenAI and Anthropic. The fear was that sophisticated autonomous AI agents, powered by these LLMs, would manage customer relationships from inception to completion, effectively sidelining Salesforce. Misinformed critics went so far as to demote Salesforce from its robust position as the central command center of a business to a mere passive database, occasionally queried by these all-powerful AI agents. The erroneous presumption was that the AI models themselves would capture all the value, relegating Salesforce to an interchangeable commodity, if not rendering it entirely redundant. Consequently, Salesforce stock, down approximately 20% this year and 40% from its peak, has been priced for precisely this faulty diagnosis, reflecting deep investor skepticism.
However, these confused critics fundamentally misinterpret the dynamic. Salesforce is not what’s being commoditized; rather, it’s the LLMs themselves that are increasingly becoming interchangeable. In this new, AI-driven world, data is the ultimate moat, and Salesforce possesses an unparalleled repository of it. As astute analysts at Wells Fargo astutely observed, “lower cost of intelligence increases value of incumbent data.” This statement encapsulates the core truth: the more accessible and powerful AI models become, the more critical the quality, volume, and proprietary nature of the data they operate on becomes.
At its core, an AI agent’s effectiveness is directly proportional to the quality and depth of the data it processes. An AI agent tasked with closing a sale, for instance, still requires a centralized location to research customer histories, log new interactions, store contracts, and customize terms. Crucially, it needs access to decades of accumulated customer data and historical context to accurately understand the nuances of each interaction and make informed decisions. This is precisely where Salesforce excels, serving as the ultimate, trusted repository of customer data for countless enterprises worldwide.
Despite the analysts’ initial delusions, Salesforce’s operational reality reflects its undeniable position as an emerging AI winner. The company has processed an astonishing 216 trillion customer records this year alone, and that number continues to climb rapidly. This immense trove of customer data—encompassing key contacts, deal histories, support tickets, marketing interactions, and comprehensive historical records—already resides within Salesforce for virtually every major company. The logistical and proprietary challenges of ripping out this data and storing it solely within an LLM are immense, not to mention the inherent risks of trusting a nascent LLM with proprietary and sensitive customer information. Clean, unified, and trusted data, underpinned by robust security and confidentiality protections far surpassing those of generic LLMs, is exactly what AI agents need to function effectively at scale, and Salesforce provides it better than anyone.
The tangible results further underscore Salesforce’s strong position. Agentforce, Salesforce’s innovative AI agent platform, has experienced explosive growth, rocketing from $100 million to an impressive $1.5 billion in annual recurring revenue within just 18 months of its launch. This rapid adoption is evidenced by over 30,000 Agentforce deals already closed, bolstered by exciting new partnerships, such as the integration of Anthropic’s Claude as a premier agent within the Agentforce ecosystem. Critically, these AI agents are not replacing Salesforce; they are, in fact, generating an exponentially increasing volume of new data that conveniently needs to be stored and managed within the Salesforce platform. This creates a powerful virtuous cycle, with Salesforce ingesting 104 trillion records last quarter alone—double that of the preceding quarter.
An often-underestimated strategic move, the acquisition of Slack, has also proven extraordinarily important to Salesforce’s AI future. Slack, serving as a dynamic collaboration hub where key decisions are debated and refined, provides AI agents with critical context and human insights that would be impossible to glean from static database fields alone. This synergistic relationship explains why Slack is experiencing record-rapid growth, recently delivering its fastest quarterly Net New Annual Order Value (NNAOV) growth since its acquisition. Slackbot users have grown over 150% quarter-over-quarter, and when Salesforce opened Slack up to outside AI agents, a million users plugged in within a single month. Visionary founder and CEO Marc Benioff’s decision to execute the largest share repurchase in company history earlier this year—a staggering $25 billion in a single quarter, representing roughly a fifth of its market capitalization—now appears incredibly astute, solidifying the position of the planet’s largest repository of customer data.
This fundamental shift in the balance of power, with Salesforce gaining, not losing, pricing power, explains why leading frontier LLM providers like Anthropic are actively seeking partnerships with Salesforce. The debut of “Claudeforce,” a pioneering partnership allowing full integration of Claude within Salesforce, exemplifies this trend. This collaboration is a win-win, designed to increase usage of both platforms and drive customers towards higher-end, premier subscription plans, further cementing Salesforce’s indispensable role in the AI enterprise stack.
Booking Holdings: The Indispensable Transaction Layer
The bearish AI narrative for Booking Holdings was similarly deceptively simple—and equally flawed. Earlier this year, some analysts prematurely presumed that if a traveler could simply ask a chatbot for hotel or flight recommendations, the need for an intermediary like Booking.com would evaporate. However, time has thoroughly debunked these skeptics. Booking Holdings’ stock has not only recovered but has bounced back to near all-time highs, a trajectory mirrored by other online travel agency (OTA) rivals such as Expedia. The critical mistake these misguided critics made was in perceiving Booking Holdings as merely a search engine. In reality, it is a highly differentiated travel transaction platform, fortified by a strong competitive moat.
The crucial distinction, often overlooked, lies between the "top of the travel funnel," where trips are discovered and initial ideas are formed, and the "bottom of the funnel," where money changes hands, and the actual planning and execution of a trip occur. Travelers are indeed increasingly turning to AI for recommendations, a trend that is genuinely ominous for metasearch and referral businesses whose primary function was comparison. However, this trend is not ominous for a company that acts as the merchant of record for approximately three-quarters of its bookings—a share that has risen four points in the past year and continues to climb rapidly. Booking Holdings excels at settling more than 100 payment methods across 50 currencies, and, crucially, adjudicating the thorny disputes and complex last-minute cancellations that AI platforms have shown absolutely no appetite to touch.
Indeed, even Google’s own leadership has explicitly stated that the company has “no intention of becoming an OTA” and harbors zero interest in acting as the merchant of record. OpenAI, too, reached a similar conclusion the hard way, retreating from its in-chat checkout functionality this spring after a badly botched rollout was widely panned by users. The complexities of handling real-world transactions, customer service, and liability proved to be a significant barrier that general-purpose AI models are ill-equipped to manage.
Furthermore, contrary to popular perception, the vast majority of Booking.com’s room nights—to the tune of 90% of all bookings—originate from independent properties and smaller hotels, rather than large, established hotel chains. These smaller properties lack the sophisticated infrastructure to manage global payment processing, multi-currency settlement, and complex dispute resolution on their own, even if they were somehow able to surface independently through AI searches. This critical operational gap is precisely what Booking’s robust infrastructure fills, forming the bedrock of its value proposition. It’s why Booking’s hotel partners are fiercely loyal and are not contemplating a departure anytime soon. The importance of this structural advantage, shielding against LLM disruption, can also be seen in Airbnb’s stock performance, which is up 40% year-to-date, partly because its inventory of exclusive properties is perceived as a strong moat against generic AI competition.
Despite their prior misjudgments as the SaaSpocalypse "death" narrative faded, these same bears have now pivoted to arguing that, much like with Salesforce, Booking Holdings will experience diminished pricing power and increased customer acquisition costs, leading to compressed margins and hindered profitability. However, this margin compression thesis is equally flawed. Accelerating AI changes will, in fact, only increase Booking Holdings’ relative power in the marketplace and make its value proposition more singular and irreplaceable.
Simply put, Booking Holdings is exceptionally well-positioned to leverage AI to gain even more market share from its less tech-savvy competitors. With Booking Holdings’ moat secure as the travel infrastructure provider of choice, there is every reason to believe that AI will drive greater traffic towards Booking’s unique platform in the years ahead, not less. We are still in the nascent stages of this pivot; AI-driven traffic has been remarkably limited for OTAs thus far. On its August earnings call, Booking disclosed that traffic sourced from large language models remains well below 1% of room nights, with no material change over recent quarters, while direct traffic held steady in the mid-60% range and grew in absolute terms. This indicates that while AI is evolving, its direct impact on booking channels is still marginal, providing Booking ample time to adapt and integrate.
But in a future where AI inevitably drives a greater share of discovery, Booking Holdings stands out. Building on its two decades of experience bidding for web browser search traffic, Booking is uniquely positioned among its competitors to apply those hard-won lessons to bidding for preferential AI traffic and advertising placements. It possesses the same tried-and-true machinery for converting a click from any source—whether from AI or a traditional search engine—into a direct, repeated, and loyal Booking customer. This singular playbook, perfected under the continued leadership of Booking’s widely admired CEO, Glenn Fogel—who is regarded as one of the best capital allocators of our era—represents a unique advantage that positions Booking to be the biggest AI beneficiary among its competitors.
IBM: The Architect of AI Transition
IBM bears wrongly believed they had stumbled onto a goldmine last month when IBM stock plummeted 25% in a single day, marking the worst performance in the company’s storied history. This sharp decline was attributed to several large clients redirecting their capital budgets towards memory investments amidst a severe global memory crunch. Although a significant portion—a third—of those supposedly lost deals ended up closing within the next few weeks, and IBM CEO Arvind Krishna earned widespread plaudits for his honest transparency regarding the market challenges, the damage to sentiment was done.
Nonetheless, a common misguided bearish narrative persists: that AI is poised to disrupt not only IBM’s $21 billion consulting business but also its hugely profitable legacy software business, which underpins the core systems of countless banks, insurers, and airlines globally. The fear is that AI, particularly in its automated and generative forms, would make human consultants less necessary and render traditional enterprise software stacks obsolete.
However, what many critics fail to grasp is that AI has, in fact, been a significant boon for IBM’s consulting business. AI now accounts for a staggering half of all new consulting signings and represents one of the largest components of IBM’s burgeoning backlog. Crucially, these AI-driven consulting engagements are conducted at far higher margins than traditional consulting services. This is thanks to IBM’s evolving billing model, which increasingly allows them to charge on the basis of outcomes and productivity gains rather than simply brute hours worked. This shift aligns perfectly with the value proposition of AI, which is to deliver tangible, measurable improvements.
Furthermore, Red Hat, the foundational software that enables a company’s AI agents to run seamlessly across any cloud and any platform, grew by an impressive 11%. Paradoxically, the advent and integration of AI are not diminishing the need for software but rather creating new demands for robust, flexible, and secure software platforms that can manage and orchestrate these advanced AI agents. This phenomenon ensures continued revenue growth in IBM’s subscription software business.
Simply put, IBM is being paid to build and facilitate the AI transition for its enterprise clients, rather than being run over by it. This strategic positioning is why IBM’s overall AI business has more than doubled over the last year, demonstrating a profound pivot and successful adaptation to the new technological paradigm.
Paranoia and Panic Are Different
Everyone—ourselves included—concedes that AI is a powerful disruptive force impacting legacy technology and software companies. However, financial markets, in their haste, appear to be "tossing out the baby with the bathwater." They are overlooking vital software companies that own indispensable assets without which AI agents simply cannot function effectively. The key question now for investors is whether a company still controls something that AI agents fundamentally need, and where the ultimate power in the marketplace truly lies. We firmly believe that this power is rapidly shifting back to software firms that, just months ago, were prematurely labeled as the biggest losers but are now quickly transforming into some of the biggest beneficiaries of AI.
Salesforce, with its unparalleled repository of proprietary customer data, owns the intelligence that AI agents cannot operate without. Booking Holdings, as the critical transaction platform, owns the operational infrastructure that agents cannot execute complex travel bookings without. IBM, through its consulting and Red Hat offerings, owns the underlying technological infrastructure that AI agents run on and are built upon. Each of these represents a differentiated and defensible moat, which is becoming even more valuable in a world increasingly driven by AI, not less. These are just three particularly compelling examples among several prominent software companies poised to benefit significantly from AI, with ServiceNow, under the capable and experienced leadership of CEO Bill McDermott, and Snowflake also standing out as core examples of companies with strong AI-resistant or AI-leveraging moats.
As legendary Intel CEO Andy Grove famously quipped, “only the paranoid survive.” But paranoia and panic are fundamentally different states. While a healthy paranoia drives necessary adaptation and strategic evolution, widespread panic across markets has led to a commendable prudence devolving into reckless lack of discrimination in discerning AI winners and losers in the software space. Software bears are missing the profound and transformative shifts taking place before our very eyes, as these resilient and strategically positioned software firms evolve into some of the biggest beneficiaries of the artificial intelligence revolution.
