OpenAI's IPO delay: Oracle and other hardware stocks fell, software stocks surged.

OpenAI's financial difficulties trigger market revaluation, software stocks surge while chip stocks face pressure.

On June 25, the New York Times reported that OpenAI may delay its IPO plans due to financial pressure, prompting the market to reassess the extent of AI's impact on traditional software companies.

The software sector saw a broad rally on Friday, with previously most threatened stocks like ServiceNow and Workday leading the gains, while Oracle, deeply tied to OpenAI, bucked the trend and fell.

RBC Capital Markets analyst Rishi Jaluria said that while market sentiment toward the software industry remains largely negative, "the most pessimistic moment may have passed." He also noted that the claim "companies will fully replace existing software solutions with AI" does not align with reality.

Software stocks broadly rally, former "hardest-hit areas" lead gains

The most prominent gains on Friday came from software companies previously considered most vulnerable to AI disruption.

ServiceNow and Workday both rose over 9%, ranking among the top gainers in the S&P 500 that day.

(Top five gainers among S&P 500 components)

Figma and Datadog closed up over 10% and 8% respectively; Adobe and Salesforce each rose about 5%; Atlassian also gained over 5%.

Raymond James analyst Adam Tindle noted that while daily stock price fluctuations are hard to attribute precisely, Friday's biggest gainers were precisely the stocks the market had been most worried about being eroded by AI, including ServiceNow and Atlassian.

Oracle bucked the trend and fell, cloud infrastructure business becomes a drag

In stark contrast to the broad software rally, Oracle fell about 3% on Friday, making it a clear outlier in the sector.

Morningstar analyst Luke Yang said the news of OpenAI's IPO delay was the main factor affecting Oracle's stock performance.

Yang explained that this news is a "positive" for software applications but a "negative" for cloud infrastructure.

Oracle has a cloud computing cooperation agreement with OpenAI worth up to $300 billion, making part of Oracle's business prospects highly correlated with OpenAI's success or failure.

Looking at Oracle's own business structure, cloud computing has become increasingly prominent. In fiscal year 2026, Oracle's cloud business revenue grew 39% year-over-year to $34 billion, while software revenue edged down 1% to $24.5 billion.

Cloud infrastructure companies under pressure, computing power suppliers face risk of downward revisions

Dragged down by news of OpenAI's financial troubles, emerging cloud infrastructure companies were also unable to escape.

CoreWeave and Nebius Group both fell on Friday, closing down about 2% and 6% respectively.

Luke Yang pointed out that both companies are highly dependent on AI infrastructure demand. Once market expectations for OpenAI's growth prospects become more conservative, the negative impact on related cloud computing power suppliers will be more direct.

Eric Jhonsa, investment manager at Dutch Asset, said on social media that the market currently prices a lot of long-term terminal value risk into most SaaS companies, while for some AI infrastructure targets, almost no long-term risk is factored in.

He further noted that progress in frontier models like GLM 5.2, export control policies, and the trend of enterprises shifting to smaller, cheaper models to reduce token costs all indicate that AI computing capital expenditures also face uncertainty over longer time horizons.

Valuation divergence intensifies, AI infrastructure investment logic faces recalibration

This market movement also pushed the valuation divergence within AI-related stocks into the spotlight.

Eric Jhonsa noted that Nvidia currently trades at about 20 times forward earnings, Broadcom about 23 times, while Marvell is at 58 times and Astera at 116 times, with the latter two having higher stock-based compensation as a percentage of earnings and free cash flow.

He believes this divergence is "absurd" and faces clear mean reversion pressure.

Meanwhile, some SaaS companies, while maintaining double-digit annualized recurring revenue and free cash flow growth, have seen their valuations compress to 10-15 times forward free cash flow (excluding stock-based compensation), which he believes provides a good margin of safety.

In terms of investment strategy, Eric Jhonsa said he is constructive on AI infrastructure targets whose valuations can be supported by capital expenditure remaining strong through 2028 (rather than extending to 2030 or beyond); for targets requiring a longer capex cycle to justify valuations, he is cautious or bearish.

He added that buying AI infrastructure targets at 15-20 times expected 2028 EPS makes him far more comfortable than paying over 30 times.

On a macro level, Eric Jhonsa said that if inflation heats up again in the second half of the year, combined with the Fed's slow response under political pressure, long-term interest rates could rise sharply. This would be the biggest macro threat to current tech stock valuations.