‘The scholar, the pragmatist, and the rainmaker:’ How three Bloomberg Beta VCs bet on AI before it became a buzzword

• Bloomberg Beta is a $375 million early stage fund housed within Bloomberg LP.
• The fund’s three partners have been prolific investors in AI and machine learning for more than a decade.
• Bloomberg Beta has embodied the what-you-see-is-what-you-get style of its leader Roy Bahat.

Roy Bahat’s LinkedIn profile reads like a laundry list of failures. In 2003, he proposed an ambitious plan in which the city of New York would trade airports to the Port Authority in exchange for the World Trade Center, an effort that “stalled under its own weight.” There was the time he was in charge of international outreach for New York City’s bid to host the 2012 Olympics, and his most notable accomplishment was apparently when he “hired a guy to knock holes in the office wall with a sledgehammer, so I could communicate more easily with my team.”

He chooses to highlight the 18 months he spent on a failed management buy-out of the business, his “struggles to retain key leadership,” and how he “spent way too much time and effort on a project to create a new website serving female gamers, before realizing we had no idea how to do it” when describing his 5-year stint as president of video game news website IGN.

Bahat refers to it as his CV of failures, and it stands out for its candor on a platform like LinkedIn, where everyone has polished their image to a mirror finish.

“We so rarely get something real,” Bahat says, referencing his LinkedIn profile but also describing his life philosophy. “In this moment where everything is being observed and discussed, I thought, how can you have something that is just fucking true?”

Bahat applies the same radical candor to Bloomberg Beta, the early-stage venture capital fund housed within the larger Bloomberg empire that he has led for the past decade. (He once published a list of reasons why people should not work with his fund.)

And, while corporate venture capital funds are not typically known for being daring or contrarian thinkers, Bahat and his two partners, James Cham and Karin Klein, have made a name for themselves by being years ahead of the curve when it comes to AI.

They made their first artificial intelligence investment in 2013, in a company called Newsle, which was later acquired by LinkedIn for a reported $52 million. In 2014, they made AI one of the primary focuses of their fund, led by Shivon Zillis, who would later work on Elon Musk’s Neuralink project and Tesla’s self-driving car. Bloomberg Beta was investing in generative AI as early as 2015, before the term “generative AI” existed.

“It felt like we were just waiting for the right moment,” Cham, a Bloomberg Beta partner and de-facto chief technologist, said, “where the rest of the world would see it.” And it appears that the rest of the world has as well. Since the release of ChatGPT last November, artificial intelligence has become the venture world’s buzzword du jour. Despite sluggish investment in other areas, money has been pouring into startups with the word “AI” in their names.

“It’s a little annoying,” Bahat said of the AI hype, “but ultimately, we recognize that fashions are just fashions, and they’ll come in and out.” We’ve been doing the same thing for ten years, and the bet is that it will pay off.”

‘I don’t even like VCs’

Bloomberg Beta was founded in 2013 when Dan Doctoroff, a former Deputy Mayor under Mike Bloomberg and CEO of Bloomberg LP at the time, approached Bahat about establishing a venture fund within the company. Doctoroff says that during his time in government, he was impressed by Bahat’s chutzpah and creativity and thought he was ideal for the job. Bahat, who had spent two years in Silicon Valley building a video game startup, was not particularly fond of venture capitalists at the time.

“I said, you guys shouldn’t do that, it’s a bad idea, corporate VCs are terrible, and by the way, I don’t even like VCs,” Bahat explained.

If he was going to do it, Bahat had specific plans for how he wanted it done. It would be a single early-stage fund tasked with identifying promising new technologies. It would have a single LP, Bloomberg, and all profits would go to Bloomberg Philanthropies. He would avoid financial services, and Bloomberg clients were off limits, but he could compete with Bloomberg as much as he wanted. He insisted on a focus on real returns and an open relationship with the founders. Bahat eventually talked himself into taking the job.

“I think at the end of the day, I didn’t like complaining about VCs without seeing if I could beat them,” he explains.

Bloomberg Beta closed its first $75 million fund in 2013, and since then has raised three additional $75 million early stage funds and one opportunity fund for follow-ons, for a total of $375 million, all with a single LP. Over the last decade, it has been one of the first investors in a number of unicorn startups, including Flexport, Replit, and Newfront.

Beta, like most venture funds, does not release detailed performance data, but Bahat tells Insider that the fund has already returned Bloomberg’s initial $75 million investment and has grown the portfolio’s value several times over. Doctoroff has only said that Beta has “done extraordinarily well” in terms of returns.

‘No bullshit’

Throughout it all, the fund has staked its reputation on a firm belief that what you see is what you get.

“There’s an art term in venture, they say he’s just being slippery,” Bahat explained, “and it implies that some degree of being slippery is okay, maybe even good.” That doesn’t work for me; if a founder dislikes directness, we’re not their people.”

Bloomberg Beta made the early decision to publish its entire operating manual on GitHub, which included sensitive information such as average check size and investment criteria. The firm opposes the use of so-called “exploding offers,” in which a founder is required to accept a term sheet within a certain time frame (Bahat claims a founder once kept an offer open for seven months).

The fund also allows any partner to say yes to an initial investment without going through a committee, and all handshake deals are considered legally binding. According to Bahat, this ensures that founders are always speaking with a decision maker and can trust them.

“I loved it because there was no bullshit,” said Tade Oyerinde, founder of education startup Campus. Campus’ pre-seed round was led by Bloomberg Beta in 2016, and the company has since raised tens of millions of dollars from investors such as OpenAI CEO Sam Altman and Peter Thiel’s Founder’s Fund.

“Most VCs are like playing poker; they don’t want you to read their minds, but Roy didn’t have a problem getting visibly excited,” Oyerinde explained.

Kieran Snyder, the founder of generative AI startup Textio, in which Bloomberg Beta invested in 2015, has a slang term for the roles of Beta’s three partners. Cham is known as “the scholar,” a die-hard technologist who has been obsessed with software since discovering the Apple Lisa in fifth grade.

He reads the academic literature and is comfortable discussing the technology behind a large language model or machine learning algorithm.

Klein, a veteran New York investor who launched SoftBank’s first east coast fund and has seen it all from high-flying decacorn to spectacular implosion, refers to herself as “the pragmatist.”

“Karen can see every angle for why something might not work,” says Bahat.

Snyder refers to Bahat as “the rainmaker,” someone who sees the vision, knows exactly how to sell it, and has a knack for making life-changing introductions. “Roy is like, do you want to meet the Treasury Secretary?” “I’ll get that arranged,” Snyder said.

Though Bahat is known for his abrasive demeanor (he says people frequently tell him he’s “very East Coast, which in California is code for asshole”), it’s important not to misinterpret his candor for arrogance. Those who work with him say he is the first to admit error. In fact, when Zilis first proposed leaning into artificial intelligence in 2013, he dismissed it as ridiculous.

Today, he is proud to say that he was completely wrong. “I want to set up a system where the system is brilliant, but I can be stupid and it still works,” Bahat explained.”That willingness to be vulnerable makes it easier to trust him,” Snyder said, noting that “in Silicon Valley, so much of the culture is posture culture.” 99.99% of the time, you’ll just hear about how people are crushing it and how they’re up and to the right. Roy is not afraid to tell the truth.”