In 2011, Marc Andreessen wrote that software is eating the world. The catchy title conveyed the transformative role software was playing across a range of industries, the thesis aged remarkably well. Today, it is the software vendors who find themselves on the precipice. In under a generation the world has pivoted, it is their turn to receive a Schumpeterian lesson in creative destruction.
Or is it? This week’s selling has been indiscriminate. This is less rotation than panic. The FT reports that the ARM CEO is referring to this as a “micro hysteria”. Well, quite. Software or SaaS has no single business model; yet the market has reacted as though every software company has the same vulnerabilities:
- Seat based commercial model(s)
- No proprietary content moat(s)
- Their market and available wallet share reducing to accommodate new, disruptive AI centric competitors.
- AI only empowering the disruptors, not the incumbents.
- Management teams lacking the tools, vision or wherewithal to adjust.
I expect you to die, Mr. Bond.
The market thesis seems simple; the SaaS industry lies prone before an AI steamroller. Or perhaps a large laser. The market is assuming no reaction from these large firms, lying supine. That there is little they can do to react, or escape.
Undoubtedly the risk profiles of the incumbent software players has increased in the last 12 months. This is true in our information services or financial data vertical, where I’ll focus.
- Growth of financial services headcount seems likely to stall, even reverse.
- New tools are arriving on the scene, with ample funding, at a dizzying pace. The investment in AI infrastructure is unprecedented.
- Incumbent reactions have been underwhelming, all hat, no cattle. Complacency? Maybe.
All these things are true. If you wanted only cherry pick the negative components of the current landscape, you’re welcome to.
It is not that simple.
Let’s review the landscape at a high level.
Kernels of truth
Let’s consider CRM. I am not a CRM expert but understand that money for upgrades might today be allocated to AI. There is no content moat. The moat, a deep one, is the sunk implementation cost. The product is sticky but growth will decelerate as budgets are reallocated. So goes the bear case, I don’t have the background to evaluate that.
In any case, I do not see a read across to financial data franchises. Their business models are not reliant on upgrade cycles. Claude or ChatGPT do not do market monitoring. Nor performance reporting. Nor trading integration. An AI platform requires the incumbents for most of the critical content.
Even with all the right content, would institutional players, or their regulators, be confident? We know probabilistic AI is far from perfect. We talked through desktop strategy in an earlier post.
Yet… AI is certainly exposing long-standing deficiencies.
Turns out our AI POC is not just a chatbot
Headcount reduction is real
Headcount will reduce in many administrative or clerical roles. I am not sure who will mourn the automation of payments, settlement or reconciliation tasks. AI can already be the first line helpdesk. It will draft your RFP response. It feels a lot like a continuation of digital transformation. Who mourned for the teller, replaced with the convenience of online banking? Who would rather return to speaking to a human to place a trade? To order a pizza?
For now, the creative disruption and loss of “old economy” headcount is clear. New roles (and industries) will only appear over time. The Economist recently noted similar angst accompanied the arrival of the PC. That’s unsurprising, what is left of mainstream media lives on hype and fear, cautious optimism sells few headlines.
It feels superficial, even cursory to just assume these companies will fail to adapt. Fail to innovate. Fail to see the need to either embrace the tech or go on an acquisition spree to recapture lost ground. Sure, tangible progress to date looks insufficient.
I worry not only on AI adoption but also on content innovation. This. isthe real moat of the financial data industry. Many companies underinvest in their content moats (I would note better M&A from S&P, Morningstar and AlphaSense. For. the industry overall, It feels overdue to reinvigorate their content strategy, finding new content to drop into LLMs or SLMs. I would be rushing to build new analytics to minimize commodification of their rapidly atrophying content franchises.
Arguably lazy, certainly inconsistent.
But, let’s flip the analysis. Software and data franchises have their risks. What happens if we turn that same microscope around? What. if public markets or private credit truly assessed the business model risk in these AI firms? Not the bull case. The base case. Plenty of PE leaders have talked about a rigorous assessment of software risk.
Yet where is the comparable rigor in assessing how the AI firms will monetize? Proof we’ll meekly accept an enormous new line item in our budget. For. themath to work AI spending will need a big B2C penetration. Without this, predator can become prey easily enough. No one ever said market narratives have to be internally consistent. Still, I marvel at this brutal localized selling, surrounded by an ocean of mid to late-stage irrational exuberance. FOMO over here, fear over there.
Final thoughts
A couple of final point to consider: firstly, the increased automation and adoption of quantitative techniques across markets. Data scientists and AI engineers are in, traditional analysts and salespeople out. Data costs rise, headcount costs fall.
For those with data centric businesses that’s a net positive. Cost of sales falls: it takes fewer salespeople in this environment as well as fewer support staff to train users. The marginal cost of content distribution is far lower than a new desktop. That’s a margin enhancing change in the business mix.
Finally, as I mentioned earlier, no allowance is being made for the quality of the management team. Do they have experience in a disruptive environment, do they understand the needs of their customers? Do they know how to innovate in a workflow or content sense? Can they construct an outcome where AI is a revenue accelerant, not a risk?
Diagnose. Strategize. Implement.
Management has always been a tough factor to quantify. It is the single most common request I heard in years of client engagement. There is consistent customer appetite for improved analytics. I can’t think of an investment factor that has been less quantified. A quick aside, Marvin Labs get a shout out for taking a look.
There are some simple proxies to management quality, how well have they used M&A, anticipating changes in their environment? Morningstar acquiring Pitchbook. S&P acquiring Kensho. MSCI’s gradual build out of private markets indices. All indicative of teams that had a conviction view of how the market was evolving. Internal capital allocation is harder to see externally, unless you use the product.
Those skills feel awfully important to me in a world where there are so few certainties. Classic VUCA. When a competitors can empower a disruptor and your competitive landscape turns upside down overnight. When you consider the volatility in the policy landscape. The obvious challenges of navigating AI as it evolves in real time. So, when you search for bargains among the wreckage, I would be judicious and thoughtful. Most of these firms have strong cashflow and the ability to buy their way out of emerging threats. Look for more: back the team with a track record of, you guessed it, knowing where that puck will be.
Leave a Reply