Microsoft prepares for its post-OpenAI era

Welcome to Runtime! Today: Microsoft outlines its plan for becoming a frontier AI model company, OpenAI's CFO walks back a suggestion that the U.S. government should pay its infrastructure bills, and the latest moves in enterprise tech.

(Please forward this email to a friend or colleague! And if it was forwarded to you, sign up here to get Runtime each week.)

Wicked smaht

Microsoft enjoyed an early advantage in the generative AI boom not thanks to anything cooked up in its AI research labs or product-development offices, but through the foresight to hitch its wagon to OpenAI. Once mutually beneficial, that relationship has clearly run its course, which means Microsoft's AI teams are going to have to walk on their own at some point.

Microsoft's Mustafa Suleyman set out the company's plans to develop something he called "humanist superintelligence" in a blog post published Thursday. "Humanist superintelligence offers an alternative vision anchored on both a non-negotiable human-centrism and a commitment to accelerating technological innovation… but in that order," he wrote.

Microsoft hired Suleyman last year in one of those weird AI-era acqui-hire deals, in which Suleyman, chief scientist Karén Simonyan, and several other key employees at Inflection AI agreed to join the company.
At the time Microsoft said Suleyman and his Microsoft AI team would be working on the company's consumer AI efforts, but Thursday's manifesto revealed the creation of the MAI Superintelligence Team, which has a wider mandate to develop core AI models.
"We want it to be the world’s best place to research and build AI, bar none," Suleyman wrote, later adding "as a platform of platforms, this is core to Microsoft’s mission of enabling others to create and invent at global scale."

Under its previous agreement with OpenAI Microsoft had agreed not to pursue "artificial general intelligence," a concept that no one on the planet can define but, nevertheless, has long been the goal of OpenAI's research. In describing Microsoft's plans for "humanist superintelligence" — which is a totally different thing, apparently — Suleyman did not exactly mask his efforts to distinguish the plans for the MAI Superintelligence team from OpenAI's work.

"We are not building an ill-defined and ethereal superintelligence; we are building a practical technology explicitly designed only to serve humanity," he wrote.
In an interview with the Wall Street Journal, Suleyman rejected the idea of AI chatbots that draw humans into a parasocial relationship, which has ruined the lives of several ChatGPT users and even pushed some to commit suicide.
"We want to create types of systems that are aligned to human values by default. That means they are not designed to exceed and escape human control," he told the Journal.

However, for all the lofty rhetoric in Suleyman's blog post Microsoft is clearly determined to play with the big kids when it comes to developing core AI models around which it can base products and services without having to pay an OpenAI or Anthropic tax. It doesn't matter whether you call those models evidence of AGI or humanist superintelligence; they are first and foremost commercial engines.

Suleyman said Microsoft will be working on creating an "AI companion," so that "everyone who wants one will have a perfect and cheap AI companion helping you learn, act, be productive and feel supported." [emphasis added]
The MAI Superintelligence team will also build models that will tackle health care and energy generation, which will allow the world to "have cheap and abundant renewable generation and storage before 2040," he predicted.

"At Microsoft AI, we believe humans matter more than AI," Suleyman wrote, and you know we've gotten to a really weird place in the generative AI boom when one of the top executives at one of the biggest companies in the world feels compelled to articulate that sentiment.

Show me the money

OpenAI CFO Sarah Friar caused quite a stir Wednesday after she made comments at a Wall Street Journal event suggesting that OpenAI was interested in having the federal government "backstop" the financing needed to fulfill its eyebrow-raising infrastructure spending commitments. After every capitalist with Jensen Huang's phone number freaked out, she walked those comments back in a LinkedIn post, saying "OpenAI is not seeking a government backstop for our infrastructure commitments. I used the word 'backstop' and it muddied the point."

Apparently that wasn't enough to tamp down speculation that OpenAI needed the federal government to help it achieve either AGI or humanist superintelligence, which prompted CEO Sam Altman to post a more detailed explanation on X Thursday. "We believe that governments should not pick winners or losers, and that taxpayers should not bail out companies that make bad business decisions or otherwise lose in the market," he wrote in the post.

The whole affair is really giving off "my 'we don't need help finding the money' t-shirt has people asking a lot of questions already answered by the shirt" vibes. Altman said that OpenAI is on track to record $20 billion in revenue this year, but in order to hit the infrastructure spending targets it has set it will need to quintuple that revenue by the end of 2027.

Enterprise moves

Chirantan “CJ” Desai is the new CEO of MongoDB, replacing longtime CEO Dev Ittycheria, who will remain on the database company's board of directors.

Joe Kauffman is the new president and chief financial officer of Deel, joining the payroll and HR software company after a decade at Credit Karma.

Josh Fecteau is the new chief data and AI officer at Teradata, following six years in data leadership roles at the company.

Noel Hamill is the new chief marketing officer at Expereo, joining the networking company after marketing leadership roles at parcelLab and PG Forsta.

The Runtime roundup

Datadog crushed Wall Street's expectations for revenue and profit and raised guidance, which sent its shares up 23% on Thursday.

Freshworks also beat estimates and raised guidance, suggesting that CEO Dennis Woodside's strategy of taking on ServiceNow directly is paying off.

HubSpot, on the other hand, watched the price of its stock fall 15% Thursday after lowering its guidance for the current quarter amid slowing growth for its marketing software.

Cybersecurity professionals have been waiting for malicious hackers to start using AI in earnest, and Google Threat Intelligence Group reported Wednesday that it "has identified cybercriminal use of 'just-in-time' AI which employs large language models (LLMs) on the fly to create malicious scripts and functions, and to obfuscate code," according to CSO.

Thanks for reading — see you Saturday!