When it comes to AI, a lot of capital is moving.

I’ve written about this a few times within the last couple of months, trying to read the tea leaves on where funding is streaming into the AI market. One place to get information is from the tech media, where analysts present their findings. Stanford HAI, for example, shows off a detailed report on the estimated $252 billion going into AI in 2024, with some geopolitical context as well. The report states :

“U.S. private AI investment hit $109.1 billion in 2024, nearly 12 times higher than China’s $9.3 billion and 24 times the U.K.’s $4.5 billion. The gap is even more pronounced in generative AI, where U.S. investment exceeded the combined total of that of China and the European Union plus U.K. by $25.4 billion, up from a $21.8 billion gap in 2023.”

Another way to analyze this is to look for the “big ones,” not only the firms that have already jumped to the top of the pile, but others that may soon follow.

“The best example is the story of a once relatively unknown chipmaker, NVIDIA,” writes Kevin Voigt, in an article that suggests likely contenders for value like Broadcom and Texas Instruments. “Valued at just over $10 billion in early 2015, the company had just crossed the threshold as a Large Cap stock. In 2025, NVIDIA became the first company to cross a market value of $5 trillion, leap-frogging tech titans like Apple and Microsoft thanks to the AI wave.”

Another place to get information is conferences and events specifically around AI and its context.

We had an event here April 9-10, an Imagination in Action summit which I help run, and MIT people and others convened to talk about the big issues. A panel discussed allocation of capital: Lisa Dolan, my business associate at Link Ventures, interviewed Guiseppe Stuto, founder of Offscript, Lisa Huang of Fidelity Investments, Libby Wayman of Liberty Ventures, Mark Machin of Intrepid Growth Partners.

Machin spoke to the days of finding out what AI could do, in the early twenty-teens, as manager of a pension fund.

“Geoffrey Hinton legendarily won that competition,” Machin said, of 2012’s ImageNet contest. “Fei Fei Li provided the data set. Geoffrey Hinton’s team from Toronto actually won the competition, and not by a little bit, by a lot. And that's when people said, ‘Oh, this is not a wasteland, a waste of time. This is real.’ And then, when I took over to run the fund, it was simultaneous with AlphaGo and DeepMind winning the AlphaGo competition, which woke the whole world up to the power of AI.”

Huang talked about seeing an Nvidia demo that brought certain points home, a 2018 presentation on using AI to QA documents.

“I think what made me double down - you know, AI in finance is still very nascent, I would say,” she said. “And we had the advantage that we were doing AI for a long time, I like to say, before anybody cared, especially as it relates to finance.”

She outlined Fidelity’s related operations:

“We do buy a lot of technology,” Huang added, “and then we also have a VC arm that invests in technology as well. We have an incubator, in which we also incubate AI technology. The business is diversified in that way.”

“This is a super-exciting moment at the intersection of energy and AI, and climate change and AI,” Wayman said, responding to a question about the challenge of balancing climate work with commerce. “We've been investing in AI applied to our fields, really, since the beginning of our fund, and many of those companies have gone on to be multi-billion-dollar companies.”

Noting the use of AI in metals recycling, and citing domestic material supply, Wayman analyzed the energy footprint of AI.

“AI is now a huge energy consumer, and we have not experienced load growth in the United States in decades,” she said, noting that an initial 4% of electrical load consumed by compute has grown by about 13-15% annually. “It’s really a pretty small, marginal growth that is just straining the system like we haven't seen in decades, and that growth is only predicted to increase.”

Later, Dolan asked Stuto about how a smaller VC fund can get an edge over a bigger player like A16z.

“Typically, we're dealing with much younger, first time builders,” Stuto explained. “Not everyone is always ready to start a company, although they have the right DNA, the right psychology in terms of how they think about the world, why they're thinking of building, what they're building, they're surrounded by exceptional talent, and then, you know, they also are quite commercial in their own regard.”

That, he suggested, feeds into strategy.

“It’s not always clear and obvious when the right time is to kind of turn into a business,” he said. “So I think it requires being patient, and not just kind of giving a founder a ton of upfront equity you're throwing on the cap table right away, especially in an age where you can get going a lot faster with AI tooling.”

Stuto also mentioned something I’ve heard a lot about, the commodification of compute, which ultimately has its own impact.

“Intelligence is becoming commoditized,” he said. “Founders are recognizing this, and they're leading with agency. And when you're leading with agency, and I'm seeing the best founders do this, I'm seeing the best founders early on value their equity differently.”

Near the end of the panel, Dolan gave every member a chance to talk a little more about success vectors for investors.

“It’s really about exceptional humans that are driving the progress of the company and will literally, well, maybe figuratively, run through walls to get things done,” Wayman said, suggesting that VCs should look for “a little fissure in the opportunity” and also adding this analysis:

“As these really large companies get larger and larger, it is impacting the earlier stage of company formation, and how people think about investing, and that's also why you're seeing these more platform-oriented funds, because there's just more value to accrue in private markets by holding longer.”

“I do think there are very specialized, verticalized applications where you really have to live it and breathe it for decades, to be able to do it,” Huang added.

“I think the key question for everybody, at the top of food chain, or wherever is the defensibility of all the businesses everybody's building,” Machin said. “And, you know, what is truly going to be sustained defensibility. So you actually have long-term value. So you actually have a 5, 10 year, 15 year future, rather than three years, and then flame out, or less.”

Listening to these folks and others talk, it seems apparent that business needs guidance, especially in terms of moving capital, as the AI market continues to heat up. How will companies set a course for the future? Probably amid a context of quite a bit of uncertainty, as what some call an “intolerable abundance” roils markets. Stay tuned for more out of our events and conferences, and some of the newest analysis of a rapidly changing world.