Just Super | E+ | Getty Images
“AI is not a place where money makes more money,” Calkins told CNBC in an interview at his London bureau on Tuesday.
Calkins was referring to high-profile deals that companies like Microsoft and Amazon have agreed to with ambitious and fast-growing foundational AI model makers like OpenAI and Anthropic.
Microsoft is investing a total of $13 billion in OpenAI, and the deal will see Microsoft acquire a stake in OpenAI and the latter adding its GPT language models to the Redmond, Wash.-based tech giant's Azure cloud computing platform. I will do it.
Microsoft struck a similar deal with Mistral, acquiring a 15 million euro ($16 million) stake in the French AI company.
In the case of OpenAI, Microsoft has non-voting observers on its board of directors.
This comes after a shocking series of events last year that led to the temporary ousting of OpenAI CEO Sam Altman, followed by hundreds of OpenAI employees being replaced by Altman. He was reinstated by threatening a coup d'état to get Microsoft to join him.
Separately, Amazon has made a huge $4 billion investment in American AI company Anthropic, which is developing the Claude AI system. Amazon owns a minority stake in Anthropic, but does not have a seat on the company's board of directors.
Google has also committed billions of dollars to Anthropic, agreeing to invest up to $2 billion last year.
British regulators are assessing whether a deal agreed by Microsoft and Amazon with a foundational AI model startup constitutes an effective merger that could lead to a substantial reduction in competition.
Microsoft denies that the contracts with OpenAI and Mistral and the hires from Inflection constitute a merger. Amazon said its partnership with Anthropic represents a limited corporate investment rather than a merger.
This is a market for smart people. The fact that you have enough money to buy and buy things like Anthropic and Mistral is great. But AI may not be a “winner-takes-all” market.
matt calkins
Appian CEO
For Calkins, whether or not these deals qualify as mergers that threaten competition in the AI field, there will be room for innovators to thrive.
He criticized the US tech giant's $500 million acquisition of British AI research institute DeepMind, saying: “If the Coalition had won the AI race, Google would have won by now.”
On the contrary, Calkins argues that he believes Google could lose out to Microsoft early on in generative AI, potentially upending the structure of Google's search business.
This follows the debacle in which Google's text-image generator Gemini introduced inaccuracies in historical photos that went viral online. Google has paused image generation of people to improve its tools. CEO Sundar Pichai called the debacle “unacceptable,” according to an internal memo obtained by CNBC in February.
Google did not immediately respond to a request for comment, but CNBC contacted them.
“This is a market for smart people,” Calkins said. “The fact that you have enough money to buy or buy Anthropic and/or Mistral is great. But AI may not be a 'winner-takes-all' market. ”
“There will be different AI algorithms for different purposes, and their value will be more or less depending on whether and how you load your own data.” added.
Calkins said the only way for AI systems to become truly smart and useful is to help them understand what we want from them to use in our daily lives.
“The best AI is the one that puts your data in it, not who bought the biggest stack,” he said.
Calkins said today's AI race is less about how smart AI actually is and more about “how much data it can eat.”
Calkins said big tech companies are “doing everything they can to get as much data as possible.” “But the game was almost over,” he added.
Because there are no specific laws preventing Big Tech from swallowing data to prevent privacy violations, these companies were allowed to obtain the data they needed to train their models.
Calkins said he is disappointed in the lack of progress in U.S. AI regulation at the federal level.
He told CNBC's “Squawk Box Europe” on Tuesday that in some ways Europe has a “head start” on AI because “the regulations are becoming clearer.”
“In the U.S., it's not so clear-cut, in part because the government is a little too friendly to big tech companies,” he said.
In March, the European Union formally approved the AI Act, the first comprehensive law governing artificial intelligence.
Calkins said companies need to be clear on how to use AI safely and ensure things like protection of intellectual property and users' personal privacy.
“There's natural skepticism in Europe…there's regulations here that push back against big tech companies in the U.S.,” Calkins said.
“I suggest that the time has come for fair use of copyrighted information. We need a clear playing field and an understanding of what data use is allowed. need to do it.”
