Providing Capital to Tech Companies: Masters In Business with Ann Berry

Reality Defender

• Ann Berry · ThreadneedleBullish

  Reality Defender is a leader in deepfake detection, with big corporate partners and rising importance.

  Berry says she invested in Reality Defender, a deepfake detection company, before deepfakes became headline news. She argues it has done well: it has signed large corporate partnerships, Gartner named it a leader among emerging companies tackling this space, and she thinks the problem is becoming more important. The company has been around about six years. Her point is supported by commercial traction and third-party recognition rather than a price or growth forecast.

  “It's done incredibly well. It signed some big corporate partnerships. Gartner identified it as the leader in terms of the emerging technology companies looking to tackle this space. And it's just getting more and more important.”at 33:37

AMI

• Ann Berry · ThreadneedleBullish

  Bullish on world models for robotics; she has invested in Yann LeCun's AMI.

  Large language models are trained mostly on text, while world models aim to understand the three-dimensional physical world, which matters for robotics. Ann says she looked actively for world-model investments and invested in LeCun's recent startup AMI, having backed the group behind this raise in 2014 before AI became mainstream. She sees real opportunity here.

  “I was lucky that several years ago, I sort of heard enough about world models that I was actually looking actively for an opportunity to start investing in them.”at 37:06

PayPal Holdings

Across the shows: 2 speakers in other episodes, 0 bullish and 2 bearish calls, past 30 days

• Ann Berry · ThreadneedleBearish

  PayPal's scale and data advantage should have produced faster AI gains; she's baffled it hasn't.

  Ann says PayPal has scale and an early advantage, including a large history of transaction data, but has not moved fast enough to use AI on that data. Barry adds that Stripe, Square and BNPL have leapfrogged it and that success can anchor a company, citing the innovator's dilemma. Ann argues that companies with success can use it to leapfrog others, but PayPal has not done so.

  “It baffles my mind that PayPal with its scale and its first view of advantage has not done more faster”at 39:07

Walmart

Across the shows: 18 speakers in other episodes, 15 bullish and 3 bearish calls, past 30 days

• Ann Berry · ThreadneedleBullish

  Walmart's data-driven ad business is a hidden gem; the company now calls itself a tech company.

  In January 2022 Ann wrote that Walmart's filings showed a growing advertising business built on e-commerce data, as brands sought alternatives to costly social-media campaigns. She says the ad business now generates nearly $6.5 billion in revenue. Walmart moved its listing to Nasdaq to signal it sees itself as a tech company, which Ann finds interesting since she used to think of it as a logistics company.

  “But it had woken up to the fact that it had a gem inside of its massive business, which was data.”at 40:53

Vody

• Ann Berry · ThreadneedleBullish

  Vody, which she invests in, fixes poor e-commerce search results by working with retailers.

  Ann tested a major e-commerce platform's search for a red sofa and found the results very poor. Vody's founder built the company to work with retailers to improve their platforms so consumers get relevant answers rather than ad-driven results. Ann invests in Vody, so her view is partly promotional, but she grounds it in a concrete search example.

  “And the quality of the feed, the search results that came back was just really, really poor.”at 46:02

AI bubble & tech valuations

• Barry Ritholtz · Ritholtz Wealth Management

  Unlike dot-com wisps, today's AI funding backs real businesses with revenue, though misallocation will occur.

  Barry compares AI to the Internet bubble, where capital funded barely formed ideas. He says today's AI companies have scale, revenue and sometimes profits, and that funding goes to real things with real potential. He expects some misallocation, as always with tech innovation, and calls the shift an industrial-revolution-scale change happening much faster. He concedes the $2 trillion market caps are still debatable.

  “But a lot of things got funded during the Internet. It wasn't just that they didn't work out. They were barely wisps of an idea. It seems that what's getting funded are real things with real potential and in many cases, revenue.”at 55:20

• Ann Berry · Threadneedle

  Tangible AI market caps rest on hard-to-measure returns; the valuation math is difficult, though the tech will transform.

  Ann argues that AI company valuations, which are tangible market caps, are tied to returns that are not yet tangible or measurable. She believes the technology will be transformational. Her concern is how transformational it becomes and when, and that is where the valuation math becomes very hard to unpack.

  “These are tangible numbers being tied to nontangible, at least right now returns. We know that there is gonna be an outcome that's tangible. It's the measurability that we're not able to see.”at 55:43

• Ann Berry · Threadneedle

  Big tech acquirers now prioritise being in the game over valuation discipline.

  Ann asks big tech CEOs and CFOs about acquisition criteria. They cite technology fit, product adjacency, team and culture, and price only comes up when she asks. These companies were historically known for discipline, with clear return-on-investment hurdles. Now they say being in the game matters more than valuing the game, which signals looser pricing.

  “And they're now saying being in the game is more important than valuing the game.”at 1:03:55

Consumer AI products

• Ann Berry · Threadneedle

  Consumers would welcome AI agents that find them the best deal, unlike ads and dynamic pricing.

  Ann distinguishes dynamic pricing, which she calls price discrimination, from an AI agent that works for the consumer. If an agent searched for the highest-yielding place to park cash, she says consumers would not push back against it, so AI could be proven to help consumers. By contrast, ads that are not serving what the user wants are a different problem.

  “I don't think you would have a backlash against that use of AI. ...So that's a case where AI could be proven to help the consumer.”at 45:01

• Ann Berry · Threadneedle

  Retailers and brands must learn to market to shopping agents as they once marketed to consumers.

  At a retail conference in Mexico City, the Liverpool CEO said consumer products companies and retailers must figure out how to market to agents, just as they have marketed to consumers. If an agent buys on a consumer's behalf, producers need to ensure the agent quickly finds their product. Ann expects a shift in how retailers and consumer product companies think about being found, using the data they control and relay to shopping agents.

  “consumer products companies and retailers are going... Have to figure out how to market to agents in the same way that they've been figuring out how to market to consumers.”at 49:09

Venture capital & startups

• Ann Berry · Threadneedle

  Early-stage valuations show discipline, unlike the 2021-22 NFT and Web3 frenzy.

  Ann recalls that in 2021-22 NFT and Web3 companies got crazy valuations. She says she is not seeing that lack of discipline in current early-stage deals, and she takes comfort from this when investing at the early stage. The discipline contrasts with the bubble-like valuations she sees in parts of public AI.

  “The one place I am seeing discipline is actually with these earlier stage companies. ...I'm not seeing that lack of discipline right now”at 56:17

• Ann Berry · Threadneedle

  Corporate startup stakes are a smart M&A pipeline, but their markups are one-time earnings bumps.

  Some of the largest US venture investors are now corporates investing their balance sheets in startups. NVIDIA has a big stake in SpaceX, and Nvidia and Salesforce have deployed billions into startups in their ecosystems, even on cost basis. Ann calls it a smart M&A tactic, since it tracks a potential target as it grows. But the markups that inflated earnings ahead of AI IPOs are one-time bumps, not ongoing profit.

  “some of the biggest venture capital firms in The United States now are corporates that are investing their balance sheets into startups.”at 1:04:09

Private equity

• Ann Berry · Threadneedle

  Higher exit valuations can no longer be expected without real fundamental improvements in the business.

  Berry argues that financial engineering alone no longer creates returns; that era has been gone for a long time. Every deal needs a clear value-creation plan with operational steps, such as growing revenue, cutting overhead, raising margins, pricing power and adjacent markets. In her view, buyers can't expect a higher exit valuation unless the business has fundamentally improved to justify it. She draws on her own operational experience, including running a portfolio company during the pandemic, to support this.

  “I don't think we're at a point anymore, Barry, that higher valuations at exit can be expected... Unless you've actually done something to the business fundamentally to justify the change in that valuation outlook”at 35:05

Enterprise AI adoption & AI software

• Ann Berry · Threadneedle

  Many firms sit on untapped data in on-premise systems; big companies have not capitalized on it.

  Ann says many businesses hold huge troves of data on their own hardware, in on-premise ERP systems that have not moved to the cloud. She is frustrated that not all big companies have capitalized on this. She sees room for nimble startups to use AI to build niche products and to be acquired by larger players.

  “so many businesses are sitting on huge troves of data on their hardware. Not everyone's in the cloud.”at 38:44

AI regulation & safety

• Barry Ritholtz · Ritholtz Wealth Management

  Expects states like California and New York to outlaw supermarket dynamic pricing based on personal data.

  Barry describes dynamic pricing that uses personal data, such as a shopper's credit score or what is already in the cart, applied to everyday items like eggs. He bets that some states will outlaw it in supermarkets, starting with California and New York, and then spread. He ties this to consumer-finance protections he believes were weakened.

  “I'm gonna bet that some states, probably California, New York, and then it spreads, are gonna outlaw dynamic pricing in supermarkets.”at 46:48

US inflation

• Barry Ritholtz · Ritholtz Wealth Management

  AI and license-plate-reader pricing could add 30-40% to goods prices unless legislation or consumer pushback stops it.

  Barry says that if AI and license-plate readers are used for pricing, they could add 30-40% to the price of goods. He cites dynamic pricing based on zip code and similar signals, such as Uber's price differences. Only legislation or a consumer arms race against AI would prevent this, so the inflation effect is conditional.

  “If you think inflation is bad now, wait till these license plate readers and AI maxing out. It's gonna add 30, 40% to the price of goods unless there's either legislation”at 47:43

AI: overall outlook

• Ann Berry · Threadneedle

  AI is transformational; the debate is how transformational and how soon, not whether it matters.

  Ann says the bubble and circular-financing concerns are legitimate, but she does believe in AI's promise. She sees it across the spectrum from early-stage innovators to the most scaled companies. Her doubts concern the magnitude and timing of returns, not whether the technology will change things.

  “is not because I don't believe in the promise of AI. I absolutely do because I see it in the early stages, the most innovative to the most scaled.”at 53:25

AI infrastructure & data centers

• Ann Berry · Threadneedle

  Big AI capex lacks rigor; the return expected on these CapEx dollars is unclear.

  Trained by a private-capital focus on return on investment, Ann reviewed earnings transcripts, calls, presentations and analyst research on AI spending. She found little rigor in quantifying the real return expected on AI capex. She calls the Internet-bubble comparison a poor one because today's companies are real and have cash-flow paths, yet the measurable return is still missing.

  “there is just not a ton of rigor at the moment in being able to wrap one's arms around what is the real return that's being expected to these CapEx dollars.”at 53:47

Open-source AI models

• Ann Berry · Threadneedle

  Chinese models like DeepSeek are rapidly adopted outside the US and Europe because they are free and cheaper.

  Ann says the US and Europe receive technologies differently from much of the rest of the world. Chinese EVs are everywhere in Mexico City and parts of Africa, and DeepSeek and other Chinese models are rapidly being adopted in regions outside Europe and the US. She attributes this partly to being free, and says even the paid versions are much cheaper than those from OpenAI, Anthropic and Google.

  “If you go to parts of the world that are not Europe or The United States, DeepSeek and the Chinese models are getting rapidly adopted.”at 58:18

Frontier AI labs & model race

• Ann Berry · Threadneedle

  OpenAI and Anthropic's fundraising war chests let them buy any startup that could contend.

  Ann argues that promising AI companies are being scooped up and acquired quickly, rather than leapfrogging the incumbents. She calls this a flight to scale, with well-funded labs like OpenAI and Anthropic using their capital to acquire anything that could be a competitor, especially in the enterprise.

  “The ability of the OpenAI's and the Anthropics to raise capital, because at the end of the day, there's almost a flight to scale.”at 1:00:29

Antitrust & tech regulation

• Barry Ritholtz · Ritholtz Wealth Management

  Big Tech concentration came from decades of acquisitions, unchecked by antitrust enforcement since Reagan.

  Barry argues the concentration in the Magnificent Seven is not just seven companies but a thousand, because the big firms have been acquiring for twenty-five years. Antitrust enforcement that would have prevented this has been absent since the Reagan era. He sees this as a reason the next generation of innovation is being absorbed by the giants.

  “It's not seven companies. It's a thousand companies. They've been doing acquisitions for twenty five years. Really, we haven't seen the sort of antitrust enforcement since Reagan was in office that would have prevented all that happening.”at 1:01:17

IPOs, M&A & deal activity

• Ann Berry · Threadneedle

  Startup exits via acquisition look robust; cash-rich corporates want to buy more than IPOs.

  Ann finds it exciting that the exit opportunity for startups feels robust, and that an acquisition is more likely than an IPO. Large corporations still have big cash balances, despite buybacks and dividend hikes, and want to buy. Startups that are nimble and build niche AI products make attractive acquisition targets for bigger players.

  “I find it exciting because it means your exit opportunity feels at the moment to be pretty robust. Not an IPO more likely than acquisition.”at 1:02:07

Consumer & retail

• Ann Berry · Threadneedle

  Creator audiences do not reliably convert to purchases, so consumer brands need content strategies.

  Ann says an audience does not necessarily translate into consumers, citing Scarlett Johansson's beauty brand shutting down as an example where celebrity and fandom did not produce consistent purchases. Some audiences do convert, though, so when she invests in consumer businesses she checks their content strategy, including founder storytelling and creator collaborations.

  “an audience does not necessarily translate directly into a consumer, by the way.”at 1:07:10