Thursday brought a volatile trading session for software equities, as quarterly earnings reports underscored the difficulty of identifying clear winners among AI-powered companies. Despite surpassing both revenue and profit expectations, Datadog (DDOG) and Figma each saw their share prices fall by more than 15%.

Profitability concerns were the primary driver behind the sell-offs. Datadog reported an adjusted gross profit margin of 80%, narrowly missing the Wall Street forecast of 80.7%. Although the discrepancy seems small, it fueled investor unease about AI's potential to erode established software profit structures. Software companies have historically commanded premium valuations built on the assumption of sustained high margins and scalable revenue growth, making even modest margin compression a sensitive signal for investors evaluating whether AI integration reinforces or undermines that economic model.

Figma faced a similar dynamic. The design collaboration platform beat earnings expectations but warned that rising AI inference costs are squeezing its margins. The company currently absorbs the computational costs of processing customer AI prompts for beta features without charging usage fees. This cost structure represents a broader industry challenge: as software vendors embed generative AI capabilities into their products, the variable compute costs associated with running inference models create a fundamentally different expense profile than traditional cloud-based SaaS delivery, where marginal costs per additional user are minimal.

"We do not charge our customers for their usage of products that are currently in beta, and we bear the cost of inference without offsetting consumption revenue," said Figma CFO Praveer Melwani on the company's earnings call.

Divergent Performance Across Software Companies

The earnings season yielded mixed results across the software landscape. Palantir (PLTR) surged 30% on Tuesday after reporting strong revenue growth. During its analyst call, the company directly challenged OpenAI and Anthropic, questioning their enterprise AI capabilities. Palantir's commercial model differs structurally from many peers under pressure: its contracts with government and large enterprise clients tend to embed AI capabilities into long-running, high-value deployments rather than layering them into self-service subscription tiers where inference costs scale with individual user activity.

Cloudflare and JFrog both posted encouraging quarterly results and recorded weekly gains. These companies framed artificial intelligence as a growth catalyst rather than a threat to their businesses, positioning themselves within the AI infrastructure stack where demand for content delivery, security, and software supply chain management scales alongside broader AI adoption.

HubSpot and Salesforce were among other notable decliners during the period.

Software Sector Outlook

The iShares Expanded Tech-Software Sector ETF fell more than 2% on Thursday. While the fund has rebounded over 35% from its April trough, it remains approximately 3% below its starting point for the year.

By comparison, the PHLX Semiconductor Index has fallen 15% since early July. Software stocks have shown relative strength against chipmakers over this compressed period, though a significant performance gap persists over longer timeframes.

The AI investment theme faced further headwinds from forced liquidations at Situational Awareness, an AI-focused hedge fund. These mandatory sales intensified pressure on infrastructure-related stocks during the summer months and fueled speculation about a value shift toward application-layer companies.

However, the current round of earnings disclosures suggests this narrative is not uniform. Even Alphabet has reorganized its AI research division following a period of industry acclaim for its development approach.

Investors anticipating a broad-based software sector recovery are confronting the challenge of timing. The sharp contrast between winners such as Palantir and underperformers like Datadog illustrates that this segment remains highly uncertain in the current environment, with outcomes hinging not merely on whether a company is "AI-related" but on the specific economics of how each business monetizes intelligent features relative to the compute costs they generate.