Software developers are under fire once again. Which companies in the sector are worth investing in?

This week, IBM acknowledged that customers have begun spending more on semiconductors at the expense of its products, leading to a new sell-off among software companies. Over the past year, the gap in market performance between the semiconductor sector and software manufacturers has reached historic levels. But since late June, the first signs of a turnaround have emerged. How will investors behave going forward, and which companies are currently a safe bet in the software sector? Yegor Tolmachev, senior analyst at Freedom Broker, provides the answers.

Over the past year, the Philadelphia Semiconductor Index—which includes 30 semiconductor companies, such as Nvidia, TSMC, and AMR—has risen by 108%. During the same period, the iShares Expanded Tech-Software ETF—which holds significant positions in Palo Alto Networks, Microsoft, and Palantir Technologies—which allows investors to bet on software and cloud service providers, as well as digital platforms operating on a SaaS model—has fallen by more than 14%.

As a result, we saw a significant gap: investors were actively shifting their funds into AI infrastructure and the semiconductor sector, while technology companies—including those operating on a “software-as-a-service” model—were left out of their scope of interest.

This year has hardly been a good one for SaaS companies; they’ve already weathered several major sell-offs. The latest round was triggered by IBM, which decided to publish unusually weak preliminary quarterly results ahead of its full report and warned that customers were shifting their budgets away from its products and toward chips and AI infrastructure.

Prior to that, starting in late June, we observed the first signs of a rotation back into the software sector, which was more technically driven. Investors were gradually taking profits in semiconductor companies and increasing their positions in high-quality software companies, a trend supported by demand from institutional investors driven by large share buyback programs. Given several sharp declines, software sector stocks are trading at low valuations based on multiples. And in the absence of any truly major successes in monetizing AI, investors remain wary of the leading software developers of the past decade, a sentiment also reflected in weak stock performance (Microsoft, for example, has lost 17% in value since the start of the year, and IBM, following this week’s plunge, has lost 26%).

However, several companies reported improvements in the monetization of their AI products as part of their latest quarterly earnings reports. Among them is Microsoft: the annual revenue run rate for its AI business exceeded $37 billion (+123%). At ServiceNow, the number of major customers for its AI bot, Now Assist, grew by more than 130% year-over-year. At Salesforce, Agentforce’s total ARR (Annual Recurring Revenue) for Agentforce grew by 205% year-over-year to reach $1.2 billion, while the number of tokens processed increased by 152% compared to the previous quarter, exceeding 28.6 trillion.

But on the scale of the business as a whole, this is still an extremely small amount, and investors need more evidence of a successful business transformation.

All in all, the emerging rotation is not yet strongly linked to a change in the fundamental picture. How might it develop?

The IBM case could serve as a new catalyst for the entire software sector—these stocks could fall even further. The IPOs of Anthropic and OpenAI also remain a major risk for the industry; these companies will need to outline a long-term strategy for monetizing their products and are likely to impact the software segment.

We do not expect any significant capital rotation or capital inflows into software developers' securities by the end of this year.

Given the shortage of computing power and key components (chips, networking equipment, etc.), the primary pricing power currently remains with hardware and infrastructure providers. Software companies’ customers are gradually shifting from a subscription model to paying for actual usage (for example, based on the number of requests to an AI model) or for results. In the future, this will allow software developers to generate more revenue and monetize expensive AI tools more effectively. But this process is slow and painful. Judging by the experience of the last three years of AI development, the timeline for such a transition keeps getting pushed back.

This quarter, Microsoft could be a good bet as capital flows into relatively undervalued software stocks, although its performance will largely be driven by its infrastructure business and its new in-house AI models.

If we view the development of AI as a given, it leads, among other things, to three important outcomes: exponential growth in data sets, an arms race in cybersecurity, and the replacement of simple, point solutions or overly general applications.

As a result, in a vulnerable and volatile market, companies with strong positions in specific niches appear to be a safer bet in the software sector. These include:

— the developers of the Snowflake and MongoDB data platforms — this is a direct bet on the growth of data volumes in the digital economy;

— cybersecurity market leaders with a deep platform-based approach to security and a proven M&A strategy, such as Palo Alto Networks and CrowdStrike — this reflects a bet on the growing number of cyber threats;

— providers of specialized industry solutions: Veeva Systems (software for the pharmaceutical and biotech industries), Samsara (a cloud-based platform for managing physical operations in transportation and manufacturing), and ServiceTitan (SaaS for companies in the consumer and commercial services sectors). They possess deep industry expertise, offering unique solutions and full integration into business processes.

— companies that built their businesses around AI from the start and have already demonstrated results. For example, Palantir.

This is not an investment recommendation.

This article was AI-translated and verified by a human editor