Model Portfolio Update: Interpreting the Market Downturn Led by the Software Sector

Following Microsoft's (MSFT) earnings release, the software sector (including Salesforce, ServiceNow, and Workday) faced a strong sell-off, with share prices falling regardless of individual company performance.

According to Bloomberg, the iShares Expanded Tech-Software Sector ETF has fallen more than 20% year-to-date, wiping out over $1 trillion in value. The issue is that the downward pressure concentrated in the software sector is spreading to the broader information technology sector, leading to a strong market-wide decline.

The prevailing interpretation for the software sector's strong sell-off is that after Claude language model developer Anthropic announced a product specialized for legal advice over the weekend, concerns arose that the profitability of SaaS (Software as a Service) companies—which have long enjoyed high margins—will slow due to AI-driven productivity gains, and that demand will decrease as clients develop solutions in-house.

I previously discussed this inevitable shift in the software industry early last year:

However, as mentioned in the article above, interpreting this as the immediate end of the software industry is a leap. While it is true that the cost of developing basic software using AI has decreased exponentially, the risk of replacing software that companies have relied on for decades with AI-written code is far greater when it comes to guaranteeing the required level of reliability, stability, and scalability. The performance of listed software companies, where large-scale B2B transactions account for the majority of revenue, will likely not suffer a significant hit going forward.

However, purchasing managers at large corporations are now likely to demand products that offer sufficient functionality and guaranteed stability from the start. For startups without existing transaction networks that are developing products from scratch, development costs prior to market launch and marketing/sales costs post-launch will increase.

Here emerges the elephant in the room. Software companies can no longer leverage an "asset-light" business structure. In the early 2000s, the rationale for Silicon Valley VCs investing in software startups was that while initial development costs were high, operating leverage would improve drastically once the client base grew beyond a certain level, allowing for high profit margins; their hypothesis was correct. Software development costs are akin to fixed costs (capitalized as intangible assets), but once developed, the same product can be sold to numerous customers. Unlike manufacturing, there are no additional costs associated with increased sales volume, resulting in very high marginal profits (the additional profit generated when selling one more unit).

However, as entry barriers to software development have lowered recently, the supply of products in the market has increased, and the marketing and sales costs required to attract increasingly discerning customers have risen. It is no longer a structure where one simply builds a product and sells it hard; rather, it has become a "capital-intensive" structure requiring continuous costs from product planning and development to sales and maintenance. Consequently, the valuation premium that SaaS companies previously enjoyed peaked in 2021, during the pandemic-induced liquidity euphoria, and has since plummeted.

I believe the news that the software sector has entered a strong downturn due to the recent launch of Anthropic's legal advisory product is closer to an ex-post justification created to assign a reason. There have been previous launches of language models with excellent coding capabilities, and CEOs have claimed productivity improvements by developing proprietary software while laying off staff. Pointing to the legal advisory product launch as the main cause of the downturn suggests there was no other suitable news directly linked to it; therefore, I judge this indiscriminate sector-wide decline as an opportunity to capitalize on.