Good morning. As much of corporate America weighs AI costs against a dizzying array of AI models to choose from, AT&T already has a clear answer: It’s betting on open models specifically to manage those costs and keep control over the flow of its proprietary data.

“You hear the stories of the ‘token apocalypse,’ and we’re not scared of the token future,” said AT&T Chief Data and AI Officer Andy Markus. “It’s something that we feel like we can manage.” While the company still uses closed, proprietary models today, it eventually wants open models to power 70% to 80% of its total AI usage.

Switching from closed, proprietary AI models to open models has already resulted in savings of 80% to 90% for AT&T in certain applications, he said.

I wrote about why AT&T made the decision to go open in a story today, but in addition to the why there is the how. It’s not easy to pick between all the open models out there, especially when new, more capable versions are frequently being released.

For Markus, choosing between open models comes down to one primary criteria: accuracy. 

“Accuracy is what makes us move in a given direction on a model,” he said. “We need to make sure we have accurate solutions that are cost performant and give us the speed and the response time required.”

AT&T’s own OTel model, which is an open model customized with telecom-specific data, was tested in 32 different variations of model type and parameter size, Markus said. And that’s a common scenario for AT&T as it parses through all the options out there.

“We test champions against the challengers against the champion,” he said. “We test everything.”

Model accuracy is also the main driver of business value for AT&T, Markus added. The company said it has seen a fivefold in-year return on investment from its AI initiatives.

Perhaps no company has triggered more AI anxiety during the last year than cloud upstart CoreWeave. Yet the company’s shares surged after-hours yesterday as it reported revenue doubled from a year earlier.

It reported revenue of $2.58 billion for the quarter that ended in June, up from $1.21 billion in the prior-year period. Highlights from the Wall Street Journal story:

“Its shares were up 18% in the Wednesday premarket, after rising 2.4% in Tuesday’s regular hours,” Barron’s reported.

“Q2 was an exceptional quarter for CoreWeave. We outperformed our plan across the board, with the operating leverage we have been building beginning to show up clearly in our results,” co-founder and CEO Michael Intrator said on an earnings call. He attributed the results to strong demand, strong execution and efforts to deepen the technology platform.

“Our AI development services, which carry higher margins, are also being adopted by a broader set of customers than our core cloud,” Intrator said.

I spoke with Intrator several times over the course of the last few months, and you can read my column about the company here. The key to understanding CoreWeave isn’t simply its massive spending and investment, but the rules around accounting for them. As I reported in July:

Intrator said on the earnings call that the benefits of scale are beginning to show up. 

“Our scale is working in our favor and the math gets better from here. Each new deployment is landing against a much larger installed base than it was even one quarter ago,” he said. “As that base grows, each new build becomes a smaller part of the whole, while contracted revenue from existing deployments remains in place. This is how we are transforming scale into operating leverage.”

—Steven Rosenbush

• Big banks are rolling out ambitious plans to pour capital into U.S. AI and energy infrastructure, the Journal reports. This week Bank of America announced a $250 billion initiative through 2027 to fund U.S. infrastructure, including AI data centers while Morgan Stanley said it would facilitate its own $1.5 trillion initiative focusing on innovation platforms and strategic industries.

• The WSJ details Nvidia’s $500 billion chip-financing partnership with a roster of Wall Street giants including BlackRock and Goldman Sachs that will pool asset-backed capital for AI companies. The plan involves selling loans backed by Nvidia hardware to investors like pension funds and insurers, betting that surging chip demand makes for solid collateral. Critics question how long that value will hold.

• AI vibe-coding startup Lovable raised $400 million at a $13.3 billion valuation, more than double its December level, in a round co-led by Menlo Ventures and the EU’s new Scaleup Europe Fund. The Stockholm-based company is on track to generate a revenue run rate of close to $600 million by the end of this month, the WSJ reports.

• OpenAI executive Brad Lightcap is stepping down, the latest leadership change at the AI lab, the WSJ reports. Lightcap had served as operating chief of OpenAI since 2022 after joining the company in 2018, according to his LinkedIn profile. In the spring he transitioned to a role overseeing special projects. Last month Fidji Simo, OpenAI’s second in command, announced she was stepping aside after going on medical leave.

• Target named its first chief AI officer as the retailer looks to integrate the technology deeper into the shopping experience through forecasting tools, agents and partnerships with Google and OpenAI, the WSJ reports. Chandhu Nair joins from Lowe’s, where he was overseeing data and AI innovation at the hardware-store chain.

The WSJ Tech Council brings together CIOs, CTOs and CISOs advancing innovation and shaping the future. Join this trusted community where tech executives connect with peers to explore emerging trends and gain the perspective they need to stay ahead of disruption.

Request an Invitation

Follow Isabelle Bousquette on LinkedIn, Instagram, X, and TikTok for more behind the scenes on her tech and AI coverage, and lately, her contributions to the WSJ Leadership Institute’s new Executive Resilience series, where she’s profiling America’s top execs about their fitness and wellness habits.

Follow Belle Lin on LinkedIn and X for her latest reporting on enterprise technology and AI.

Steven Rosenbush is chief of the enterprise technology bureau at the WSJ Leadership Institute. He also has a column. You can follow him on LinkedIn.

Tom Loftus is the editor of The Morning Download. He suggests following Isabelle, Belle and Steve on their various social channels. But if you insist, here’s his LinkedIn.