Alphabet is merging an internal Google Research team called Brain with DeepMind, a move designed to bring two groups focused on artificial intelligence closer together as the battle for AI heats up.
Google acquired DeepMind in 2014 for a reported $500 million and has until now run it as an independent unit out of the U.K. DeepMind has been one of Alphabet’s “other bets,” performing futuristic work, such as teaching computer systems to beat top-ranked players of the Chinese board game Go.
“Combining all this talent into one focused team, backed by the computational resources of Google, will significantly accelerate our progress in AI,” Alphabet CEO Sundar Pichai said in blog post Thursday.
Jeff Dean, who currently leads Google’s AI efforts, will be promoted and given the title of chief scientist at Google, reporting to Pichai. He’ll head up the “most critical and strategic” technical projects related to AI, the first of which will be a series of powerful, multimodal AI models.
The move marks Google’s latest reorganization in response to the rapid developments in AI, following OpenAI’s launch of the chatbot ChatGPT late last year. CNBC previously reported that Google reshuffled its Assistant organization to prioritize the company’s AI chatbot Bard.
“The pace of progress is now faster than ever before,” Pichai wrote. “To ensure the bold and responsible development of general AI, we’re creating a unit that will help us build more capable systems more safely and responsibly.”
DeepMind has been able to operate separately from Google’s core research, enabling it to move quicker on breakthroughs such as AlphaFold, which can predict 3D models of protein structures. The two divisions, DeepMind and Google Research, have also reportedly had tensions in the past, leading DeepMind to seek more independence.
DeepMind CEO Demis Hassabis will lead the development of “the most capable and responsible general AI systems,” Pichai said. That research, he added, “will help power the next generation of our products and services.”
Brain, the Google Research team merging with DeepMind, is focused on AI and machine learning. Pichai said Google Research will continue work in areas such as algorithms and theory, privacy and security, quantum computing, health, and responsible Al.
In addition to the blog post, Pichai sent a lengthier memo to staffers about the changes.
James Manyika, Google’s senior vice president of technology and society, will now oversee Google Research, along with his existing teams, Pichai said. Manyika will report to Dean and the changes will take place over the next few weeks, the memo said.
Here’s the text of the memo, which CNBC obtained:
Hi everyone,
We’ve been an Al-first company since 2016 because we see Al as the most significant way to deliver on our mission. Since then, we’ve used Al to improve many of our core products, from Search, YouTube and Gmail to the incredible camera in Pixel phones. We’ve helped businesses and developers harness the power of Al via Google Cloud, and we’ve shown Al’s potential to address societal issues like health and climate change.
Along the way, we’ve been lucky to have two world-class research teams leading the entire industry forward with foundational breakthroughs that have ushered in a new era of Al.
The pace of progress is now faster than ever before. To ensure the bold and responsible development of general Al, we’re creating a unit that will help us build more capable systems more safely and responsibly.
This group, called Google DeepMind, will bring together part of Google Research (the Brain team) and DeepMind. Combining all this talent into one focused team, backed by the computational resources of Google, will significantly accelerate our progress in Al.
As CEO of the new unit, Demis Hassabis will lead the development of our most capable and responsible general Al systems — research that will help power the next generation of our products and services. Jeff Dean will take on the elevated role of Google’s Chief Scientist, reporting to me. In that capacity he’ll serve as Chief Scientist to Google Research and Google DeepMind. Jeff will help set the future direction of our Al research and head up our most critical and strategic technical projects related to Al — the first of which will be a series of powerful, multi-modal Al models.
This move brings together two leading research groups in the Al field. Their collective accomplishments in Al over the last decade span AlphaGo, Transformers, word2vec, WaveNet, AlphaFold, sequence to sequence models, distillation, deep reinforcement learning, and distributed systems and software frameworks like TensorFlow and JAX for expressing, training and deploying large scale ML models.
Google DeepMind will operate as a nimble, fast-paced unit, with clear points of connection and collaboration with Google Research and the PAs.
With this change, James Manyika will now oversee Google Research along with his existing Tech & Society teams. Many of Research’s technological advances have shaped core products and features across Alphabet and will continue to do so. Working closely with Jeff as Chief Scientist, Google Research will continue its focus on fundamental and applied research across a broad portfolio. This means cracking seemingly impossible, foundational and long-term challenges in computer science — including in Al and ML — that benefit people’s lives around the world, from algorithms and theory to privacy and security to quantum computing, health, responsible Al, and more.
We’re announcing these changes today and will take the next few weeks to get the new teams into place.
Please join me in congratulating Demis, Jeff, and James on their new roles and their continued collaboration. The Google Research and DeepMind teams have laid the foundation that brought us to this inflection point.
I’m so excited for the next phase of this journey the progress we’ make against our mission, and all the ways we’ll help people reach their potential with increasingly capable and responsible Al.
Govini, a defense tech software startup taking on the likes of Palantir, has blown past $100 million in annual recurring revenue, the company announced Friday.
“We’re growing faster than 100% in a three-year CAGR, and I expect that next year we’ll continue to do the same,” CEO Tara Murphy Dougherty told CNBC’s Morgan Brennan in an interview. With how “big this market is, we can keep growing for a long, long time, and that’s really exciting.”
CAGR stands for compound annual growth rate, a measurement of the rate of return.
The Arlington, Virginia-based company also announced a $150 million growth investment from Bain Capital. It plans to use the money to expand its team and product offering to satisfy growing security demands.
In recent years, venture capitalists have poured more money into defense tech startups like Govini to satisfy heightened national security concerns and modernize the military as global conflict ensues.
The group, which includes unicorns like Palmer Luckey’s Anduril, Shield AI and artificial intelligence beneficiary Palantir, is taking on legacy giants such as Boeing, Lockheed Martin and Northrop Grumman, that have long leaned on contracts from the Pentagon.
Read more CNBC tech news
Dougherty, who previously worked at Palantir, said she hopes the company can seize a “vertical slice” of the defense technology space.
The 14-year-old Govini has already secured a string of big wins in recent years, including an over $900-million U.S. government contract and deals with the Department of War.
Govini is known for its flagship AI software Ark, which it says can help modernize the military’s defense tech supply chain by better managing product lifecycles as military needs grow more sophisticated.
“If the United States can get this acquisition system right, it can actually be a decisive advantage for us,” Dougherty said.
Looking ahead, Dougherty told CNBC that she anticipates some setbacks from the government shutdown.
Navy customers could be particularly hard hit, and that could put the U.S. at a major disadvantage.
While the U.S. is maintaining its AI dominance, China is outpacing its shipbuilding capacity and that needs to be taken “very seriously,” she added.
The launch of OpenAI’s updated Sora 2 AI video service kicked off another round of anxiety among musicians, actors and other content creators.
Sora allows users to generate short videos for free by typing in a prompt. The app is only available on iOS devices and is limited to invitees, meaning people need a code to access it. Still, Sora has climbed to the No. 1 spot in Apple’s App Store, and OpenAI said this week it hit 1 million downloads in less than five days after launch.
CNBC’s Julia Boorstin got access to Sora 2 and tried prompts like “show me a video of a fat orange cartoon cat eating lasagna” and “create a superhero that wears a black cape and is saving a woman from a burning building.” Some of the prompts failed due to copyright infringement, while others worked. Watch the video to see what happens when we put Sora 2 to the test.
Applied Digital shares jumped 16% on Friday after the company posted strong first-quarter revenue that was boosted by artificial intelligence data center demand, putting the stock up more than 350% for the year.
Here’s how the company did compared to LSEG estimates:
Loss per share: Loss of 7 cents vs. a loss of 13 cents expected
Revenue: $64.2 million vs. $50 million expected
First quarter revenue of $64.2 million was up 84% from a year ago, when it reported $34.85 million in revenue.
The data center company reported earnings after the bell on Thursday.
During the quarter, Applied Digital built on its $7 billion lease agreement with CoreWeave that was announced in June for another 150 megawatts at the firm’s Polaris Forge 1 campus in North Dakota. The additional capacity brings the anticipated contracted lease revenue for the project up to $11 billion.
Read more CNBC tech news
“With hyperscalers expected to invest approximately $350 billion into AI deployment this year, we believe we are in a prime position to serve as the modern-day picks and shovels of the intelligence era,” CEO Wes Cummins said in a release.
The new 150 MW building will join two other data cell blocks, each hosting 100 MW and 150 MW. The company noted that one building is nearly complete and construction will begin on the other.
Applied Digital also secured funding from Macquarie Equipment Capital for a second campus in North Dakota, dubbed Polaris Forge 2. The estimated $3 billion factory will hold two 150 MW buildings, bringing the total leased capacity to 600 MW across both campuses.
An initial 200 MW of power is expected to come online in 2026 and reach full capacity in 2027, the company said.
The company had a net loss of $18.5 million in the first quarter, a loss of 7 cents per share. A year ago, the company posted a net loss of $4.29 million, a loss of 3 cents per share.
Analysts polled by LSEG expect a loss of 15 cents per share for the second quarter on revenue of $76 million.