It might read a tad counterintuitive but make no mistake, recent events are arguably the most instructive, and bullish, we have seen for some time. And it has nothing to do with geopolitics (if anything, they continue to make the investing picture murkier), the direction of interest rates, fiscal policy nor any other macro variable we keep tabs on to instruct our investing patterns.
August 3rd 2025. Remember that date – it may go down as one of the seminal moments in recent history. Why? It was the day that Sam Altman, CEO of OpenAI – owner of ChatGPT – said, “entering the fast fashion era of SaaS very soon.”
When Altman spoke of this, he was predicting that the software moats that took decades to build – like proprietary coding and complex user interfaces – would soon be replicated (or bypassed) in a matter of weeks, making software as disposable and cheap as a Zara t-shirt.
It is only now coming to dawn just how prescient that statement was. At the time, that comment marked a near-term top in the S&P 500 Software Index, and saw the sector fall 6% over the subsequent month. For an index that had outperformed the broader market by 3 ½ times over the preceding decade (and by 40ppts over the preceding 5 years), it was viewed as a self-serving, sensationalist headline that created a buying opportunity. Indeed, by the end of October, the S&P 500 Software Index had rallied back to all-time highs.
However, in a little over 3 months, the S&P 500 Software Index has halved from those all-time highs. The Australian equivalent – the S&P/ASX 200 Information Technology Index – has fared only marginally better, down 40%. Even portfolio stalwart and AI disruptor itself, Microsoft, has entered a technical bear market, down 24%.
Not only is this destruction of wealth real, targeting what were previously regarded as the highest quality companies in the world, the disruption seems to be occurring at a cadence that is accelerating, not slowing down.
This quantum and speed of shareholder wealth destruction is occurring in a manner that, on the surface, seems the antithesis of having a constructive equity market backdrop. Yet we argue that it was a necessary pre-condition for equity market returns to broaden, and for the equity bull market to lengthen. Just like a bushfire, sometimes we need to scorch the earth to let new growth take hold.
Although we suggested that Sam Altman might be viewed as prescient, he may just have been reacting to what his peers and competitors at Anthropic and Alphabet were doing at the time.
Barely three weeks before Altman made his comment, Anthropic, owner of Large Language Model (LLM), Claude, launched “Claude for Financial Services”. More than just an AI model, it was an out-of-the-box solution designed specifically for the needs of professionals working in large financial institutions - whether they’re analysts, portfolio managers, or data scientists. In essence, “Claude for Financial Services” broke the ”behind the wall” defence. Historically, your most sensitive financial data was locked in what amounted to high security digital vaults (like Snowflake, or Databricks), where AI couldn’t reach it. Anthropic didn’t break into these vaults, they basically build a “secure bridge” that allows the AI to enter, analyse the data and build complex models (like a discounted cashflow model), without the data ever leaving the firm’s secure environment. While the data is still proprietary and protected, the ability of a company to charge for “workflow layers” or analysis of the data, is now compromised (in essence, it could turn these data subscription businesses into commoditised data companies who no longer have the ability to sell value-added services on top).
Google’s release of Project Genie, an experimental research prototype that allows Google AI Ultra subscribers to create interactive (i.e., virtual) worlds from text or images, has seen $35 billion of value across the video game sector evaporate. If an AI agent can generate a life-like, interactive 3D world from a single photo, the multi-billion-dollar video and gaming platforms – which developers previously paid to manage complex tasks like physics and lighting – need to alter their value proposition (or charge customers differently).
On January 12th, 2026, Anthropic launched Claude Cowork, which may well be considered the “agentic tipping point” - a desktop application that can autonomously plan and execute multi-step tasks across a users’ local files and cloud applications (think balancing an SME’s books across thousands of accounts receivable and payable invoices, bank balances, general ledger entries etc). It can’t plug-in to the hundreds of banking institutions and payment providers like Intuit or Xero can, but it undoubtedly challenges how much those platforms will perform for customers, and their ability to upsell their own agentic capabilities.
And then in February, Claude revealed its Legal Plugin, which effectively announced the shift from being a "model provider" (selling the engine) to an "application provider" (selling the finished car). The Plugin can automate work like contract analysis and review, NDA triage (non-disclosure agreement), and compliance workflows.
While we and others believed that "behind the wall" databases would offer some protection against AI models, the reality is that the most successful firms are now using these "agentic" tools to dissolve those walls entirely. The transition to the "fast fashion era of SaaS" means that productivity is no longer locked behind high-margin, specialised software silos. Instead, as AI agents become capable of executing real-world use cases at scale - handling everything from complex tax reconciliation to legal triage - we may be witnessing the precursor to a significant spike in global corporate productivity.
This destruction of wealth in specific software sub-sectors is not an indictment of the broader market, but a reallocation of value as LLMs upend business models. Although the focus is on the US$2tn in market cap destroyed across the software sector, investors should not lose sight of the fact that global equities have added a net $3tn in value over that period. History suggests that such profound technological adoption does not just destroy old value; it creates a broader, more robust foundation for equity market returns by democratising the "intelligence" that once required an expensive, per-user subscription.
By moving past this initial "disruption" phase, investors can begin to look toward the immense productivity dividend that may eventually lift margins across a whole swathe of companies - not just those with a Silicon Valley zip code. We are seeing the early evidence of this regime change today: over the past three months, Small-Caps, Mid-Caps, and the equal-weight S&P 500 have handily outperformed the "Magnificent 7" and the market-cap weighted S&P 500.
Although three months is not a definitive trend, it is highly possible that this broadening of returns is the direct result of AI’s "agentic leap". While the disruption of world-class software businesses is undoubtedly painful for many portfolios, the irony is that it may result in a more sustainable, long-term bull market. In order for the "AI bubble" to avoid a 2000-style burst, the benefits needed to accrue beyond the providers of the technology to the users of it. It may not feel like it yet, but beneath the surface, the market is telling us that the "Productivity Dividend" may be starting to accrue.
Receive ongoing access to LGT Wealth Management’s insights and observations - a curated stream of thought leadership, market perspectives, and strategic updates designed to inform sophisticated investors.
