Masters in Business Sep 18, 2026 1h 11m 48m saved
With Glen Kacher, founder and Chief Investment Officer of Light Street Capital
Taiwan Semiconductor, Nvidia, Broadcom and AMD together made up about 40% of the public side of Light Street Capital's portfolio as of recent filings.
Most investors reading the AI trade are now arguing about when the capital spending gets cut. Glen Kacher's position is that the hyperscalers are not ahead of demand at all; they are behind it.
"So today, demand is still running way ahead of supply. And so this doomerism that has grown up around AI, in my mind, is misplaced."
Kacher worked at Julian Robertson's Tiger Management from 1993, spent 13 years at Roger McNamee's Integral Capital Partners co-leading 46 venture deals, and has run Light Street from Palo Alto since 2010. The fund was down 26% in 2021 and 54% in 2022, then up 46%, 59% and 37% in the three years that followed.
The full interview is covered here so you can skip it. 71 minutes of audio, 23 minutes of reading.
Here are the 17 predictions that matter.
Key Takeaways
Nvidia holds roughly 85% of AI accelerators, and Kacher thinks the market is underrating its ability to keep innovating
Inference will end up a larger market than training, which is why he expects many chip designs rather than one
Four chip names are about 40% of his public book, on the logic that you put capital where the innovation is
Microsoft fumbled its OpenAI lead, in his words, and is only now back in his portfolio
The real contest he watches is free open-source models against paid frontier models, not Anthropic against OpenAI
He dates the AI infrastructure cycle at 15 to 20 years and says we are a third through the first leg
The bottleneck is physical: memory and foundry capacity, and it pushes prices higher than they need to be
Data-center opposition is an education problem, and the answer is generating power behind the meter
Agents are the 2031 product, and they make a user consume 5X the tokens of someone typing prompts
A fund down 54% in 2022 then up 46%, 59% and 37% is what living with technology beta looks like
1. Peter Lynch to Tiger
Kacher read Peter Lynch in college and said that is what decided his career. He could not recall whether it was One Up on Wall Street or Beating the Street first.
The book sold him on the search itself
I was just caught by this idea of the search for great companies, great ideas.
Glen Kacher
What appealed was the detective work
And the way he told the story of finding these companies and researching them, it was really a journey of a detective trying to figure out what would matter in the future.
Glen Kacher
He reached Tiger Management at 22 through Michael Bills, a former Tiger partner who taught finance at the University of Virginia's McIntire School of Commerce. Roughly half of Tiger's investment staff had been to UVA at some point, he said, so the firm knew what to expect. He was there full-time from 1993 to 1996 and kept working for Tiger during his year at Stanford.
Robertson called him at 6 a.m. in business school
Julian would occasionally wake me up with a 6 a.m. Phone call when I was in business school.
Glen Kacher
That was 6 a.m. Pacific and 9 a.m. in New York, half an hour before the open. The names they worked through were Dell, Microsoft, Compaq and Cisco. He looked briefly at financial institutions with Rob Pitts but had already studied technology and stayed with it.
He knew within months he was in the right place
Two or three months into the job, I ended up sitting two chairs away from Bill Gates at an analyst meeting
Glen Kacher
2. Buy Only the Winner
After Stanford, Kacher spent 13 years at Integral Capital Partners, leading or co-leading 46 venture deals, among them Agile, ArcSight, Blue Nile, Epiphany, Fortify, Interwoven, LogMeIn, OpenTable and Overture. Asked what tied them together, he credited McNamee's discipline about coverage: a four- or five-person firm cannot follow an entire industry.
Focus where the change is fastest
So you have to focus in when you're investing and say, where is the change really happening most quickly? Where is it most dramatic? That disruption equals opportunity as an investor.
Glen Kacher
The host put it back to him as a career-long pattern of identifying the disruption and getting in front of it before incumbents react. Kacher added the timing caveat himself.
Early is good, too early is not
It's great to be early, but not too early, right?
Glen Kacher
His stated takeaway from the venture years is never to buy the second- or third-best company in a category.
The market splits two-thirds, a quarter, and scraps
The number one player is going to get two-thirds of the market. Number two player might get 20%, 25% tops, and everyone else fights for the scraps, right?
Glen Kacher
Asked whether winner-take-all is a technology phenomenon or a general one, Kacher said mature industries show it too, naming General Electric and Coca-Cola on distribution advantage, but that in technology the mechanism is reinvestment: more dollars going back into the product, compounding a lead at the point in a market's development where compounding matters most, and only then building the moats that keep competitors out. He also flagged the other half of the pattern, that incumbents get disrupted and that disruption tends to come from smaller companies, with AI and the chips behind it as the interesting test of that rule.
The market may be underrating Nvidia's ability to innovate
A lot of people sort of assume NVIDIA is going to lose their massive market share in AI accelerators, which is roughly 85%.
Glen Kacher
3. Training vs Inference
Asked to define terms, Kacher separated the two workloads. Training uses AI accelerator chips, which Nvidia dominates with graphics processors originally built for the mathematics of physics, lighting and shading in video games; the same mathematics turned out to suit model training. Using a finished model to answer questions or execute tasks is inference, and that can run on a simpler chip, usually with more memory and at lower cost, with different software approaches to get there.
He explained the industry shorthand: XPU crosses out the graphics and stands for whatever comes next. He named Google's TPU, Trainium and Groq as separate approaches to the same problem.
Inference is the bigger market
Because ultimately that will be a larger market than the training market.
Glen Kacher
4. What Tiger Taught Him
The host listed the people Kacher has worked with or alongside, from Robertson and McNamee to Philippe Laffont, Steve Mandel and Chip Morris, along with the firms he rattled off: Alger, Viking, Lone Pine, Impala, Matrix and Coatue. Kacher's answer was about three principles rather than technique. The first was hiring and backing only people the firm respected.
Any doubt about a CEO's integrity ended the discussion
We looked for people that we thought were of high integrity and if there was any question about the integrity of those CEOs and CFOs, we were out.
Glen Kacher
The second was that there were no shortcuts: an analyst had to explain how the conclusion was reached, that a company was well positioned and that something in its industry or product set was changing its trajectory. The third was using the firm's success to help other people. He said many of the Tiger alumni who started their own funds modelled them on what they had seen.
At Integral he worked with McNamee, John Powell and Chip Morris, all three from T. Rowe Price, in Kleiner Perkins's building, which put him near investors who backed Jeff Bezos and Google's founders early.
5. The 2010 Launch Pitch
Light Street launched in 2010, with the financial crisis still driving the news. Kacher said the pitch was that the board had been reset and multiples were low, and that four things were emerging at once: smartphones becoming the dominant platform, social media redefining media with Facebook and Twitter, cloud taking internet technology into business software, and e-commerce selling anywhere at low cost on the back end Amazon had built.
Four things at once
So there was a real emergence of these four powerful things, mobile, social, cloud, and e-commerce.
Glen Kacher
And the sharing economy needed all of them
You couldn't have had Uber and Lyft and DoorDash without having e-commerce and the mobile phone
Glen Kacher
6. Why Palo Alto Matters
The host quoted the firm's own description of itself as the Silicon Valley home team, one of the few hedge funds working at the center of the technology industry, and said his first reaction was disbelief that there were not more. Kacher agreed the number is small, named Coatue, Tiger Global and Whale Rock in Boston as evidence that it can be done elsewhere, and then made the case for geography.
Being where the innovation is centered is an advantage
There is a real advantage to living and working in the place where the innovation is centered.
Glen Kacher
He added that a fundamental innovation like AI pulls that advantage back toward Silicon Valley, because the small number of strong AI founders want to be near each other.
Asked whether San Francisco feels like the late 1990s, he moved the comparison earlier.
It feels more like the mid-1990s
I'd say more in the mid-90s, probably.
Glen Kacher
His reason is that the innovation is happening at a low level in the stack, with what he called a renaissance in hardware: semiconductors, networking, down to printed circuit boards.
Every layer has to improve at once
You have to innovate at sort of every level of the stack in order to grow at a 10x, 100x rate.
Glen Kacher
He tied the current structure back to the early 2000s, when the semiconductor industry was allowed to consolidate and given the capital to do it, naming Avago and Hock Tan as the organizers of that process. The result is a small number of companies competing for a very large market: AMD, Broadcom, Nvidia, TSMC in Taiwan on the manufacturing side, and equipment makers such as ASML.
7. The Four Chip Names
Taiwan Semiconductor, Nvidia, Broadcom and AMD were about 40% of the public portion of the portfolio as of recent filings. Kacher's justification is the same focus rule.
Nvidia's share of the GPU market
And right now, NVIDIA's got 80-plus percent market share in the network GPU market.
Glen Kacher
AMD matters, he said, because it is one of the two major players in desktop and server CPUs, which becomes more valuable as agentic AI drives the next leg of growth. Broadcom earns its place on the strength of the TPU work it has done with Google over the years, which he said has now opened doors at OpenAI and other large AI buyers.
TSMC wins whoever wins
TSMC makes the chips for all three of those companies
Glen Kacher
He described that position as close to an oligopoly or monopoly and said it is why the firm wants to own it alongside the designers.
8. Microsoft and the AI5
The host recalled that in 2024 Kacher grouped those four with Microsoft as the AI5, and said it significantly outperformed the Magnificent Seven that year. Asked whether the thesis still holds, he said Microsoft is the one that has moved in and out, mostly out.
Microsoft mishandled its OpenAI lead
The uptake of Microsoft's AI that was somewhat powered by OpenAI really didn't work that well. And that was a real miss for them.
Glen Kacher
Pressed on whether that opened the door for Anthropic, he agreed. He said Microsoft then pulled back on developing and funding its own AI work, and that he now considers it back in the game. Two things brought it back into the portfolio. The first is security.
Microsoft is the largest security company there is
Microsoft is the largest security company in the world.
Glen Kacher
And AI creates the demand for it
One of the things that we've learned is that AI creates a lot of security vulnerabilities for businesses.
Glen Kacher
The second is the repositioning of Azure, which he called impressive, alongside cutting spending in gaming that was not working.
On whether frontier AI becomes a duopoly, he said the contest between Anthropic and OpenAI is happening in real time, and declined to write off Google, which has Gemini, distribution through search, and now supplies Apple's AI efforts.
Google still has the technology and the distribution
I wouldn't count Google out and they still have great technology.
Glen Kacher
9. Open Source vs Frontier
The real fight, in Kacher's account, is not between the two frontier labs. It is between them and free models. Open weights can be downloaded at no cost and run on local hardware or on commodity hardware in someone else's data center, and they are competing directly with the more expensive frontier offerings. His read is that there is room for both.
Open models are hard to regulate by construction
You really can't regulate very well because you can install them on your own software.
Glen Kacher
He noted that users can adjust an open model to behave how they want, which raises safety questions, and that regulators have limited options because the models are already loose.
The host then asked whether security-conscious enterprises will avoid open source and default to the two frontier labs, if only because a hack is easier to defend after the fact. Kacher split the question in two: controlling AI inside the company so it cannot reach or leak sensitive data, and defending against what an outside attacker can do with open-source tools. Both are cybersecurity budget lines, which he said creates opportunities for CrowdStrike, Palo Alto Networks and Microsoft. He added a third requirement: an agent acting for a user has to honor that user's access permissions.
10. Learning From Venture
Light Street's private investments include Uber, Lyft, Slack, Pinterest, Toast, Harry's, Everlane, Box, BlackBuck and ezCater. Asked what venture teaches about public markets, Kacher said the value is in what a founder is free to do.
Founders have no incumbent customers to protect
They're starting with a blank sheet of paper.
Glen Kacher
A startup will adopt new technology more aggressively than an incumbent managing five or ten years of existing customers, so asking founders which chips and infrastructure they are using reveals demand early. In 2022 and 2023, those conversations pointed him at Nvidia, AMD, Broadcom and Marvell for the public book. The host added that Kacher had presented Palo Alto Networks at a 2018 conference at a far lower price than today's, and Farfetch at a later one.
He also told a longer story from his Integral years. Bill Joy, one of Sun Microsystems' four founders and then a partner at Kleiner Perkins, described a group of engineers, Kacher thought at Caltech, using graphics processors for early AI calculations around 2005 or 2006.
Joy's conclusion stayed with him for years
The conclusion of that team and of Bill himself, one of the great pioneers of Silicon Valley, was that GPUs would be the best chip architecture to do AI calculations.
Glen Kacher
He kept asking Nvidia about AI on subsequent visits, and Jensen Huang would discuss it, but it was a tiny end market and crypto mattered far more to the stock. Then the two moved in opposite directions at once.
Crypto crashed exactly as AI took off
It was very fortunate in the back half of 22, crypto crashed at the same time as AI was taking off.
Glen Kacher
The market was watching the wrong thing
The stock market was much more focused on what was happening with crypto that drove the stock down and not as focused on this emerging opportunity in AI.
Glen Kacher
That, he said, is when the firm built its Nvidia position.
11. Down 54%, Then Up 59%
The host read out the numbers: down 26% in 2021, down 54% in 2022, then up 46%, 59% and 37%. Kacher called 2021 and 2022 frustrating. The firm had traded 2020 well, short going into the pandemic and then long software and e-commerce through the quarantine period, and the transition out of that was difficult.
Software broke in weeks, not months
Software really got hit in 22 over the course of a month or two.
Glen Kacher
What he credits for the recovery is reassessing rather than holding on: identifying AI as the emerging category, naming the companies positioned for it, and buying them at what he thought were attractive valuations while solving for the next six to 24 months.
12. Long the Disruptor
Asked to explain being long the disruptor and short the incumbent, with long Uber against short rental-car companies as the example, Kacher pushed back on the framing.
He does not run them as pairs
We don't necessarily do paired trades
Glen Kacher
He treats the long and the short as separate opportunities, and said the press tends to simplify the market into good and bad, which is how things get overdone. His live example is software, which he said has bounced hard in the past month or two after the view took hold that AI dooms it. The objection he offered is historical.
Good incumbent technology lasts
If they solve a problem really well, they can stick around for a long time.
Glen Kacher
Banks still run mainframes
Many brokerage firms and banks are running mainframe solutions still because it works.
Glen Kacher
His mechanism is that nobody spends a new technology on an old, working, boring process; new technology gets pointed at things that create an advantage over competitors.
So the core systems stay
So those core systems don't tend to get swapped out.
Glen Kacher
The job is to trade both extremes
We're taking advantage of the doom and gloom as well as the excitement about the new things.
Glen Kacher
13. Demand Is Ahead of Supply
The host quoted Kacher on variant perception: looking for a mismatch between perception and reality, with a thesis about when and how the mismatch resolves. Kacher's answer went back to the detective framing and to a line of McNamee's about fieldwork.
McNamee's rule on legwork
Everybody goes out for a pass.
Glen Kacher
Talk to customers, suppliers and the innovators themselves, he said, which is how the firm gets the real story. The mismatch he sees now is in how AI is being described: as a job-destroying evil empire overspending its way to a crash. His counter is what users do.
It is not a productivity story to him
It's a demand story.
Glen Kacher
End users, he said, are choosing these tools in their browser, in agent software and in their development environment because they get more done, and that choice is what drives the build-out of compute capacity.
The host raised the risk of over-allocation, noting that Jensen Huang raised his estimate of AI infrastructure spending by 2030 from $1 trillion to $4 trillion, and asked when the coyote steps off the cliff. Kacher's answer was that the question mistakes where the hyperscalers are in the cycle. Amazon, Microsoft and Google have become the partners building the compute stack for Anthropic and OpenAI.
They are not running ahead of demand
And the reality is they're not ahead today. They're behind.
Glen Kacher
The over-investment has not happened
The negative doomers are expecting them to over-invest, but today that's just not happening.
Glen Kacher
The constraint, he said, is physical: memory companies, which the firm also owns, and Taiwan Semiconductor can only add capacity so fast.
Which is why he calls the doomerism misplaced
So today, demand is still running way ahead of supply. And so this doomerism that has grown up around AI, in my mind, is misplaced.
Glen Kacher
Pressed on what a bottleneck means when models keep getting better every few weeks, he gave a price answer.
The bottleneck shows up in price
I think the bottleneck drives the pricing higher.
Glen Kacher
The two of them then compared what they actually spend. The host said his firm pays $200 a month for a seat plus an enterprise arrangement, and that it has not turned out extreme. Kacher said Light Street has run its research process on software it has built over 15 years, buys data for analysis and sentiment tracking, and that none of it is cheap; for programmers, he said, spending $100 a day is not unusual. That cost, in his account, is exactly where the demand for free open-source models comes from, along with the ability to run a model on your own hardware, adjust its weights and train it on your own data.
14. A 15-Year AI Cycle
Kacher has described the build-out as a ten-year demand cycle with the bear case being capital spending cut the moment returns disappoint. Asked why a decade, he said the comparison is the shift from client-server to internet architecture.
The whole computing stack is changing
We're changing the entire stack of computing.
Glen Kacher
And these shifts come once a generation
These computing cycles happen about every, you know, 15 to 25 years
Glen Kacher
His technical point is that the old model stored data and then searched and retrieved it, while AI builds a custom answer to each question against each user's data every time, which is far more compute-intensive and far more useful. On his historical pattern, a new architecture takes about 15 years to reach a quarter of the industry's total capacity.
At least 15 to 20 years
It's at least a 15 to 20 year cycle that we're looking at.
Glen Kacher
And the first leg is a third done
We're kind of a third of the way through the first leg.
Glen Kacher
He broke the cycle into three overlapping phases: five to ten years of infrastructure, then the platform or operating system layer from roughly year six to year 16, then applications from year 11 to year 21, at which point applications become where businesses spend and where the innovation happens.
15. Data Centers and Politics
The host raised the political backlash to data centers, noting that some states have already banned them, and asked how Kacher treats that as an investment risk. Kacher said any bottleneck that slows adoption is a problem, that Light Street owns Nvidia and Taiwan Semiconductor on numbers it thinks are still better than Wall Street expects, and that the obstacle is real but not yet large enough to change the case.
His answer to the objection is economic. He grew up near Northern Virginia.
Northern Virginia is the industry's capital
That is the data center capital of the world.
Glen Kacher
Data centers, he said, have lifted local tax receipts and created demand for electricians, plumbers and construction labor, and called the ongoing tax revenue a long-running payment to those communities. He said he finds it unfortunate that some communities are not more positive about it and blamed a lack of explanation from elected officials.
On electricity prices, his answer was specific.
Generate the power behind the meter
The source of electricity needs to be behind the meter, right?
Glen Kacher
Where existing supply is short, he said, the developer has to bring its own. His live example is a data center under consideration in San Mateo, California, whose plan is to run Bloom Energy fuel cells on natural gas, which he described as producing almost no emissions and almost no audible noise.
Residents opposed it anyway
Residents have rallied against it because they've heard data centers are bad.
Glen Kacher
Asked whether the November midterms would settle it, he said the problem is not electoral.
It is an education problem
It's an education challenge.
Glen Kacher
Each project, he said, has to set out how it procures energy, how many jobs it creates, what tax revenue it generates, and whether it goes on the grid or brings its own supply, at both local and national level.
16. The Mag 7 and Apple
The host recalled that in the spring Kacher liked Amazon, Google and Nvidia and was not a fan of Meta, Tesla or Apple, and asked whether that still holds. Kacher said it does, with Microsoft moved back into the good column.
On Apple, he made a positive case conditional on execution. A phone holds both personal and business data with the security to keep them separate, is carried everywhere and is on most of the time, which puts it in a position no other device occupies for recommending actions across both halves of a user's life.
Apple's upside depends on getting AI working
If Apple can get things right, that should accelerate their opportunities or earnings over the next couple of years.
Glen Kacher
The host was blunter, calling Siri a decade-long embarrassment, and suggested the Google arrangement is a win for both sides at a cost that is trivial to Apple and pure profit to Google. Kacher agreed it is possible and said it comes down to execution, then defended the pattern as deliberate strategy rather than lateness.
Apple waited on smartphones too
They waited. They watched what Nokia did, what BlackBerry did
Glen Kacher
His qualification was that arriving second only works with a large balance sheet and users willing to wait for a better answer.
17. Agents in 2031
Asked what defines AI for consumers and businesses five years out, Kacher gave a one-word answer and then the consumption math behind it.
Agents work when you are not watching
The ability to have the technology working on problems when you're not directing it.
Glen Kacher
Which multiplies token consumption by five
It leads to users consuming 5X the tokens that you would consume just directing AI as you would a search engine.
Glen Kacher
He described the same pattern on both sides of a user's life: agents handling problems arriving in a personal inbox or family messages, and agents solving problems with coworkers on the business side.
Bonus Insights
Asked for his mentors, Kacher named Robertson for the example of running an investment business with integrity and intellectual honesty, and McNamee and John Powell at Integral for the years in his early thirties when he learned how to run a firm and was given room to both succeed and fail on private deals.
His current reading is Empires of Light, on the electrification of America and the fight between Edison, General Electric, Tesla and Westinghouse.
Edison marketed AC through the electric chair
Edison really pushed that AC was dangerous. And to the point where he promoted it for the electric chair, because it made AC look bad and dangerous.
Glen Kacher
Asked by the host whether an elephant was electrocuted to make the point, Kacher said many different animals were, and a prisoner, and that the first electric chair did not work well. He drew the parallel to AI being cast as an evil empire, and then drew the investment lesson from how the fight ended.
Westinghouse won on execution and industrialization
Westinghouse won out with, you know, steady execution and industrialization of the back end.
Glen Kacher
That is the role he sees for Amazon, Microsoft and Google in AI, and he singled out AWS.
AWS is the part of Amazon that matters
AWS is the more important part of the company.
Glen Kacher
On what he watches and reads for entertainment, he named X, because the argument about his own industry is on it, and the show Friends & Neighbors as a guilty pleasure.
His advice to graduates is to publish
You have all the tools today to make an impact.
Glen Kacher
Do the research, put it online, post it on X, and interact with people in the industry, he said. Uncover the story behind a stock and make a good recommendation and the hiring decision looks obvious.
Publishing research is a public audition
You're trying out for the world in real time.
Glen Kacher
His answer to what he wishes he had known in 1993 was about pace. He traced the smartphone from the Apple Newton, which did not work well, to General Magic, which also did not, to Palm, which got adoption without email, messaging or telephony, then the Palm Treo, then BlackBerry as the first genuinely working version, clunky and loved for its keyboard, and finally Apple.
Things move fast and still take years
While things happen fast, it also takes years for things to really develop.
Glen Kacher
Applied to AI, he said the technology will do far more in a few years, and named data security and colleagues' comfort with giving an agent access to company data as obstacles beyond the ones already discussed.
He thinks the surface has barely been scratched
We're just scratching the surface
Glen Kacher
Kacher's bottom line is that the AI trade is a supply-constrained demand story rather than an over-investment story, that the money belongs in the few companies that supply the accelerating compute, and that the cycle has 15 to 20 years to run with agents as the product that drives the next leg of consumption.
Products, Companies & Tools Mentioned
Light Street Capital (Kacher's technology-focused hedge fund and private investor, run from Palo Alto since 2010)
Nvidia (Roughly 85% of AI accelerators and 80-plus percent of the GPU market; the position was built in the back half of 2022 while the market watched crypto)
Taiwan Semiconductor (Makes the chips for Nvidia, AMD and Broadcom, which is why Kacher owns the manufacturer as well as the designers)
AMD (Coming up in accelerators and one of two major players in desktop and server CPUs, which he says matters more as agentic AI arrives)
Broadcom (Its TPU work with Google has opened opportunities with OpenAI and other large AI buyers)
Microsoft (Fumbled its OpenAI lead in Kacher's view, pulled back, and is back in the portfolio on Azure and on being the largest security company in the world)
OpenAI and Anthropic (The frontier contest he says is playing out in real time, with Anthropic the beneficiary of Microsoft's miss)
Google (Not counted out: Gemini, search distribution, its own TPU, and now supplying Apple's AI)
Amazon (The Westinghouse of the analogy; AWS, not retail, is the part Kacher says matters)
Apple (A conditional buy case: the phone holds both halves of a user's life, but Siri's record makes it execution-dependent)
Marvell (One of the four chip names that came out of asking AI startups what they were building on)
ASML (Named as one of the semiconductor capital-equipment companies with very large share and large moats)
CrowdStrike and Palo Alto Networks (Beneficiaries of the cybersecurity spending AI forces; Kacher presented Palo Alto at a 2018 conference far below today's price)
Groq (Named as another approach to inference, one of several flavors he expects)
Bloom Energy (Its natural-gas fuel cells are the behind-the-meter answer in the San Mateo data-center proposal residents opposed)
Uber and Lyft (Private investments and his example of businesses that needed mobile and e-commerce to exist at all)
Coatue, Tiger Global and Whale Rock (The other technology-focused public managers he named, all outside Silicon Valley)
Books & Resources Mentioned
One Up on Wall Street and Beating the Street – Peter Lynch (The books that made Kacher want to be an investor, for the detective framing rather than the stock picks)
Empires of Light (His current read, on Edison, Tesla and Westinghouse; he uses the campaign against AC as the analogy for how AI is being described)
Friends & Neighbors (What he streams, named alongside X as where the argument about his industry happens)
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: