IBM’s AI Spending Shock Sends Software Stocks Reeling
IBM shares fell as much as 25% on Tuesday. The drop erased roughly $70 billion in market cap. It was the company’s largest one-day decline in decades. CEO Arvind Krishna warned that IBM AI spending pressure across the enterprise is diverting budget away from traditional software and mainframe deals. Preliminary second-quarter numbers came in soft. Revenue reached $17.2 billion, and adjusted earnings per share hit $2.93. Both figures fell below Wall Street’s estimates of roughly $17.86 billion and $3.02. IBM called this its slowest revenue growth pace in more than a year.
Krishna said customers “sharply altered spending priorities during the closing weeks of June.” They directed capital toward servers, storage, and memory. The rush was driven by anticipated AI-driven price increases. Several large contracts simply failed to close within the quarter. The IBM AI spending shift rippled across the sector. It pulled down shares of Microsoft, Salesforce, Workday, SAP, Accenture, Cognizant, and Infosys. Gil Luria, head of technology research at D.A. Davidson, said the miss “may also indicate a crowding out of spend by enterprises.” Companies may be holding off on software renewals to pay for AI token consumption instead. IBM reports full second-quarter results on July 22.
Starbucks builds its own software instead of renewing
Alan Shimel’s companion piece, “A Coffee Company Just Spooked the Entire Software Industry,” connects the dots. Starbucks disclosed plans to replace a Microsoft inventory system and an IBM maintenance platform with its own AI-assisted software. This sits inside a roughly $400 million annual software bill and a broader $2 billion cost-reduction push. Shimel’s take is blunt: “AI-assisted development just made it thinkable” for enterprises to build rather than rent. However, owning the code also means owning every problem it develops long after launch. Starbucks already abandoned one AI-powered inventory tool this year after it produced unreliable results.
EU ruling strips algorithmic platforms of safe harbor
Jon Swartz’s report on the CJEU Grand Chamber ruling in WebGroup/Coyote marks a turning point for platform liability. The court held that automation does not neutralize control. When a platform’s algorithm actively determines how content is amplified for its own commercial interest, the legal shield under DSA Article 6 falls away. Basic indexing and chronological feeds stay protected. However, “For You” pages, trending modules, and engagement-driven recommendation systems do not. The ruling directly names Meta, Google, YouTube, and TikTok. Its practical reach is expected to extend well beyond the EU as global platforms rethink algorithmic design.
Chip shot: FIFA’s Connected Ball controversy
Frank Vizard’s piece on the World Cup’s “Connected Ball” sensor technology closes out the show. Adidas’ Trionda ball is fitted with a Kinexon sensor sampling at roughly 500 Hz. It is meant to settle disputed offside and touch calls. Instead, it has fueled controversy. Disputed calls include a disallowed Portugal-Croatia goal and a contested Norway-England sequence involving the overhead Spidercam wire.
Watch the full conversation with Alan Shimel, Mike Vizard, Jon Swartz, Mitch Ashley, Fred Wilmot, and Drew Gutstein on today’s episode.
