What this covers

Alex Sacerdote founded Whale Rock Capital Management, a technology-focused firm running more than $17 billion across hedge fund, long-only, and hybrid strategies. The conversation opens with his highest-conviction position, Anthropic, and uses it to walk the AI stack from chips to models to applications. Along the way he lays out how he reads technology cycles. Adoption speed depends on friction, which is the radio vs dishwasher adoption model: consumer products spread faster than ones that must plug into back-end systems. Curves can also stall, as electric vehicles did at 10 to 15 percent of cars. Even the best curve rewards only firms with an advantage, and he cites RIM, Palm, and Nokia in smartphones.

The range is wide. On the bullish side, he argues that agentic coding is the real unlock, sizing coding alone at roughly a half-trillion-dollar market, and that Anthropic has about half the compute it needs. He also describes a modified Rule of 40 for AI exposure, which multiplies AI's share of sales by category market share. His clearest example of low P/E multiples is Celestica, which he found near 8 times earnings after it became the sole supplier of Google's TPU servers.

The more contested claims concern model providers. He says frontier models are differentiated rather than commodity, and that Anthropic is building lock-in with its harness and surrounding tools. He also concedes that stalled model progress and catching-up open-source models could trigger a race to the bottom. Other threads include why large endowments underweight mega-cap tech, whether AI-native challengers can displace sticky software incumbents, and how Whale Rock uses the scuttlebutt method while leaving thesis-changing judgment to people.

Sharpest takeaway

Sacerdote argues that a single lens combining technology S-curve position, durable competitive advantage, and underappreciated earnings power lets investors find compounding winners, and that AI infrastructure is still early on its adoption curve.

  • Exponential unit growth at the right part of an S-curve turns cheap-looking P/E multiples into large earnings gains
  • Digital-era moats such as network effects, scale, standards, and IP are as durable as offline moats, and leaders in tech tend to compound
  • AI usage is still roughly 10 basis points of knowledge workers, so infrastructure demand is early and compute is short

The claims · ranked88 claims · weighted by value

This asset isn't compiled yet

You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.

0.86

Radio reached 100 percent penetration in about seven years, one of the fastest ever, while dishwashers adopted slowly because they must be plugged into back-end systems, so B2B products generally adopt slower than consumer products.

factualhigh valueestablishednovelty 3/4durability 4/4· Alex

“The radio S-curve is one of the fastest ever — it took seven years to reach 100% penetration. But the dishwasher S-curve is the opposite, because it needs to be plugged into the back end.”

0.81

The electric vehicle S-curve was expected to reach 40 to 50 percent of cars but hit a wall at 10 to 15 percent, so S-curves do not always complete and must be re-underwritten as conditions change.

factualhigh valueestablishednovelty 3/4durability 3/4· Alex

“we thought maybe 40% to 50% of cars would go electric, but it hit a big wall at 10% or 15%.”

0.80

Technologies follow S-curves, and the inflection happens when barriers to adoption are removed, as when Jobs cut the smartphone price toward $200, added touchscreen and easy use, and when Tesla got EV prices toward $40,000 and range to about 300 miles.

causalhigh valueestablishednovelty 2/4durability 4/4· Alex

“So all the barriers to adoption were eliminated. And then you rocket. When those barriers are removed, that's the tornado of demand”

0.80

Even companies in the best S-curve of all time can lose if they lack a competitive advantage, as RIM, Palm, Nokia, HTC, LG, and Motorola went from positive to zero or negative economics.

causalhigh valueestablishednovelty 2/4durability 4/4· Alex

“But if your name was RIM, Palm, Nokia, HTC, LG, Motorola — I can go on forever — zero, zero, zero, negative, negative, negative, negative.”

0.73

If AI model improvement stalls and open-source models catch up, model providers could enter a race to the bottom that hurts their stocks, though chip companies would benefit because they do not care which model wins.

forecasthigh valuecontestednovelty 3/4durability 3/4· Alex

“if good enough is good enough, I won't have a business.”

0.73

In tech the leader usually goes bigger faster and wins, because scale, big capital needs, and compounding lead keep the leader ahead, with exceptions occurring mainly during paradigm shifts such as AOL's failure to move to broadband.

causalhigh valuecontestednovelty 3/4durability 3/4· Alex

“the leader goes bigger faster, and wins. Most of the time, the leader just keeps going.”

0.69

Agentic coding is the true unlock of AI: heavy internal Anthropic usage at about $100 a day per person implies $20,000 to $30,000 a year per coder, and across roughly 20 million coders that is a half-trillion-dollar market from coding alone.

forecasthigh valuecontestednovelty 3/4durability 2/4· Alex

“people were spending $100 a day on tokens — which, if you do the math, comes out to $20,000 or $30,000 a year. And if you think about how many coders there are in the world, 20 million, you've got a half-a-trillion-dollar market just from coding alone.”

0.69

Only about 10 basis points of knowledge workers are truly using AI at depth, Anthropic may have around 14 to 15 million daily active users with only a small share doing serious work, and the infrastructure layer is roughly 10 percent penetrated, so adoption is early.

factualhigh valuecontestednovelty 3/4durability 2/4· Alex

“Sundar said it's 10 bips of the knowledge workers of the world.”

0.69

Large endowments are underweight the largest tech companies because they hold few public mega-caps, assume large caps have no alpha, and see half of public exposure as international, creating an opportunity in mega-cap tech that Whale Rock believes is mispriced.

causalhigh valuecontestednovelty 3/4durability 2/4· Alex

“We just think there's a huge structural underweight of the largest tech companies in the world.”

0.64

Older technologies are sticky and rarely killed by new platforms because mobile games did not destroy console games, tablets did not destroy PCs, and smartphones did not destroy PCs, and companies prefer to buy rather than build software.

causalhigh valuecontestednovelty 2/4durability 3/4· Alex Sacerdote

“Mobile video games didn't hurt console games, the tablet didn't hurt the PC, and the smartphone didn't hurt the PC.”

0.63

Of roughly 60 companies pursuing foundational models around 2023, nearly all startups fell away, and the field narrowed to a three-horse oligopoly of leading model providers.

factualhigh valuecontestednovelty 2/4durability 2/4· Alex

“At the time, two or three years ago, there were 60 different companies going after it.”

0.63

Within one to five years, AI-native companies could challenge each strong software incumbent, and if their data advantage can be obviated, the incumbent could be replaced with AI, though old software is sticky and enterprises prefer to buy.

forecasthigh valuecontestednovelty 2/4durability 2/4· Alex

“you could imagine a world where, in one, two, three, four, five years, you could have a brand-new AI-native company going after each one of these very strong incumbents.”

0.63

Anthropic has about half the compute it needs today, before a large additional take-up, and Marc Andreessen is confident there will not be enough compute in the next four years.

forecasthigh valuecontestednovelty 2/4durability 2/4· Alex

“So Anthropic has half of what they need right now, and that's before this huge take-up.”

0.61

Celestica, a contract manufacturer with IBM supercomputing heritage, became sole supplier of Google's TPU servers and holds 50 to 60 percent of the cloud ethernet switch market, and it was trading near 8 times earnings when Whale Rock found it.

factualhigh valuespeaker onlynovelty 4/4durability 3/4· Alex

“we noticed they were the sole supplier of the Google TPU server three years ago. The stock was trading at eight times earnings.”

0.58

For AI businesses, a modified rule of 40 multiplies the percent of sales from AI by market share in that category, so a company with 30 percent AI sales and 30 percent category share scores 60, a more useful signal than classic growth plus margin for AI exposure.

definitionhigh valuespeaker onlynovelty 4/4durability 2/4· Alex

“For AI, we have a new rule of 40: what percent of your sales are AI? 30%. And what's your market share in that category? Say 30%. You'd be 60.”

0.57

Whale Rock's research uses Philip Fisher's scuttlebutt method of talking to suppliers, customers, and competitors, and AI can write reports and quarter reviews but cannot supply the judgment about what changes the thesis.

normativehigh valuespeaker onlynovelty 3/4durability 3/4· Alex

“The AI can be a great reporter; it can't pick into the future.”

0.57

Anthropic is building a full ecosystem around its API, including an SDK, Claude for Cowork, an orchestration layer, and a harness, which parallels how AWS in 2013 invented products that built lock-in beyond commodity servers.

causalhigh valuespeaker onlynovelty 3/4durability 3/4· Alex

“They're building a whole panoply, a whole ecosystem of products around the API.”

0.57

Most investors focus on the next quarter or year and do not believe two to four years out can be predicted, yet an investor who understands the S-curve, the moats, and how to model them can forecast these outcomes.

normativehigh valuespeaker onlynovelty 3/4durability 3/4· Alex

“The world doesn't think exponentially; they're so focused on the next year, the next quarter.”

0.57

Few investors get this right because they treat each strong year at Nvidia as a bubble, and many semiconductor analysts missed the foundational model layer because they lacked the broader picture across decades of S-curves.

causalhigh valuespeaker onlynovelty 3/4durability 3/4· Alex

“every time with Nvidia over the last four years, it's like, 'Oh, they had a great year — oh my God, it's got to be a bubble.'”

0.57

Unlike cloud servers, which are largely commodity with stickiness coming from software on top, frontier models are highly differentiated by training methods and skills, such as Anthropic's strength in private equity and finance and Google's strength in ingesting PDFs, and routers that switch between models mask this differentiation.

causalhigh valuespeaker onlynovelty 3/4durability 3/4· Alex

“there's tremendous differentiation within — there are different training methods and different skills they're good at.”

0.57

AI workloads grow about 10x per year, pushing every element of data center hardware to physical limits, which de-commoditizes an industry that had no meaningful hardware innovation for years despite 25 to 40 percent annual compute growth.

causalhigh valuespeaker onlynovelty 3/4durability 3/4· Alex

“Now you go to AI, and the workloads are growing 10x every year. And they're pushing every single aspect of this hardware to the physical limits of what it can do.”

0.57

Investing at the right part of an S-curve with a strong business model produces exponential rather than linear earnings growth, so underappreciated long-term earnings power can take earnings from $1 to $10 and let investors buy great companies at very low P/E multiples, such as Nvidia at about 4 times earnings in 2023 and Tesla at about 5 times in 2019.

causalhigh valuespeaker onlynovelty 3/4durability 3/4· Alex Sacerdote

“When you get the right part of the S-curve, you get exponential unit growth.”

0.56

Whale Rock's framework combines three tests, namely a technology S-curve, a durable competitive advantage, and underappreciated long-term earnings power, because when a firm sits on the right part of an S-curve with a strong moat, earnings grow exponentially rather than linearly.

causalhigh valuespeaker onlynovelty 2/4durability 4/4· Alex

“We have an investment framework — S-curve, competitive advantage, and then underappreciated earnings power. When you get the right part of the S-curve, you get exponential unit growth.”

0.54

Investors must estimate how tall an S-curve is, not just whether it has taken off, because that determines when to sell and how long to hold, and S-curves can be dynamic and change height over time.

normativehigh valuespeaker onlynovelty 3/4durability 3/4· Alex

“It's how big is this S-curve? How tall is it — so you know when to sell, how long to hold on?”

0.54

AI servers require liquid cooling and printed circuit boards of about 40 layers versus about 10 for regular servers, which creates few qualified suppliers and turns a commodity component into a critical part whose failure takes the whole system down.

factualhigh valuespeaker onlynovelty 3/4durability 3/4· Alex

“With a regular server, you need 10 layers; with these AI servers, you need 40 layers. And there are very few PCB suppliers who can make this.”

0.54

Selling AI that performs human work requires forward-deployed engineers at the customer's side, which is a different selling motion from installing a fixed system, and incumbents may not have the people for it.

causalhigh valuespeaker onlynovelty 3/4durability 3/4· Alex

“So you need the FDEs — forward deployed engineers.”

0.54

Whale Rock initially assumed AWS would be 50 percent deflationary and the firm only 1 to 2 percent penetrated, but cloud did not deflate because building it yourself costs about the same, so the total addressable market was much larger than the $600 billion of IT systems first estimated.

factualhigh valuespeaker onlynovelty 3/4durability 3/4· Alex

“And then we said it's probably going to be 50% deflationary, and therefore we're 1% or 2% penetrated. But over time, we realized it actually wasn't deflationary.”

0.54

There is alpha in the largest companies because a small cap needs only one investor to notice, while a mega cap needs about 100 diversified portfolio managers to recognize it is a winner, so the firm can realize it before most generalists do.

causalhigh valuespeaker onlynovelty 3/4durability 3/4· Alex

“But it takes 100 people — 100 diversified PMs — to realize Google's not a loser, it's a winner.”

0.53

If one of the leading AI players falters, the compute it had committed becomes surplus, as with Oracle's canceled deal that Meta absorbed, creating overcapacity risk even though AI demand would likely absorb it elsewhere.

causalhigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“if one or two of the players falter and lose their position and can't compete, that could be a lot of compute they don't need in the future.”

0.53

Enterprise AI application penetration is under 1 percent, and because the adoption path is steep and fast, Whale Rock calls it an L-curve or backwards L-curve rather than a classic S-curve.

forecasthigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“the enterprise application AI market is less than 1% penetrated. When we talk about S-curves, we call this one an L-curve, just straight up.”

0.53

Whale Rock got into Anthropic's round by building relationships with management and the CFO, presenting a 90-page deck on the coding market built with Claude Code, and then receiving a larger allocation than expected.

factualhigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“We did a 90-page PowerPoint deck where we used Claude Code to scour the internet for all the feedback we could find about the coding market”

0.53

When a technology reaches roughly 30 to 40 percent penetration, exponential growth typically stops, the sell side catches up, and large upside surprises disappear, which marks a point to reassess holdings.

normativehigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“generally, when something gets to 30%, 40% penetrated, you stop having exponential growth, which means the sell side catches up and there are no longer big beats.”

0.53

Crowd size at events such as Gartner IT Symposium, where 30,000 American CIOs attend, and the packed AWS ballroom signaled exploding demand before it appeared in reported numbers.

factualhigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“So you could see the demand exploding before it happened.”

0.53

Anthropic's moat rests on four things: critical intellectual property that keeps its share of code high, a strong enterprise brand where CIOs say Claude first, escape velocity from fundraising and roughly 10x sales growth, and recursive improvement as its coding models feed back into its own model.

causalhigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“Go talk to any CIO, and the first thing they'll say is Claude.”

0.53

Whale Rock went from 40 to 50 percent software exposure five years ago to selling almost all of it, because AI application products were not moving the needle and customers could not be charged for them, and the firm entered 2026 net short.

factualhigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“Five years ago, we might have had 40% or 50% of our portfolio in software.”

0.53

Once Anthropic leads in coding, it can feed that coding back into its own model development, a form of recursive improvement, so its innovation pace accelerates and it may reach a lift-off stage.

causalhigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“Anthropic, now that they're leading in code, they set that code back onto their own model. It's this concept of recursive improvement.”

0.53

Incumbent software faces four pressures from AI: CIO priority shifts toward AI token spend with faster ROI, budget squeezes from that spending, reduced ability to raise prices every year, and seat losses as companies gut jobs and freeze hiring.

causalhigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“First, a lot of software companies... Second, if they're spending all that money over there, it pushes on the budget”

0.53

Because AI is accessed directly through a browser rather than requiring plumbing into infrastructure, its adoption can rise nearly straight up, a pattern Sacerdote calls a backwards L-curve, unlike the 30 to 50 percent growth typical of cloud and SaaS.

causalhigh valuespeaker onlynovelty 3/4durability 2/4· Alex Sacerdote

“what's amazing about AI is that, at least with consumers or even business, you just open up the browser and it's there.”

0.52

Digital competitive advantages such as network effects, scale, industry standard status, platform lock-in, critical intellectual property, and brand can be as durable as offline moats, and many leading tech companies combine several of them.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Alex

“What we found over the years is that some of the competitive advantages within the digital world are more powerful — if not equally or more powerful — than in the offline world.”

0.52

After ChatGPT launched in November 2022, Whale Rock chose to invest in chips and infrastructure first, because those layers get the demand first and their winners are identifiable regardless of which foundational model company wins.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Alex

“we said we wanted to be in the chips and the infrastructure first. Not only do they get the demand first, but we know who the winners are.”

0.52

Conviction is strongest when the investor, the analyst, and a respected outside investor all like a stock, forming three legs of a stool, and two-way idea sharing with other investors supports this.

normativehigh valuespeaker onlynovelty 2/4durability 3/4· Alex

“I call it the tripod — when I like something, and then my analyst likes it, and then somebody who I really respect also likes it”

0.52

Enterprise AI adoption risks being slowed by security-conscious IT departments, and adoption requires top-management evangelists, as happened with cloud when agencies like the CIA and Capital One proved cloud could be secure.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Alex

“That was one of the big things with cloud, where everybody was afraid that it's not secure to have your data in the cloud. And then we saw the CIA do it, and we saw Capital One.”

0.52

The AI application layer is not yet clear, because it is hard to see where foundational models end and applications begin and whether applications can build moats, so Whale Rock has not invested much there, though apps usually emerge three to four years after infrastructure.

normativehigh valuespeaker onlynovelty 2/4durability 3/4· Alex

“To date, we found that area to be pretty risky, because where does the foundational model end, and where does the application begin?”

0.52

AWS achieved lock-in because it treated commodity servers as a new computing paradigm, invented products that were visible to it before others, and built stickiness that kept customers in the ecosystem, which made it the first mover with a seven-year lead.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Alex

“Amazon saw it was a new way of doing computing, so they invented all these products — which they could see before everybody else — that slowly built lock-in.”

0.52

Applications historically take time to mature, as the first three or four years of the iPhone showed, so the application layer's dominance in market capitalization may come later, and proof of durability usually arrives a few years after the platform shift.

forecasthigh valuespeaker onlynovelty 2/4durability 3/4· Alex

“Well, A, it always comes later. In the first three or four years of the iPhone, the applications really took time.”

0.52

Coding is the true unlock of AI because agentic coding tools, starting with Claude Code in 2025, moved from autocomplete-style assistance to running agentically, causing the coding market to explode and making Anthropic's coding franchise central to its position.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Alex Sacerdote

“The big kicker was code, and this is the true unlock of AI.”

0.51

In April 2020, Whale Rock bought into Stripe at about a $35 billion valuation, having estimated its take rate at 40 to 50 basis points against Adyen's 25 to 30 and found TPV well above the disclosed figure, closer to $1 trillion than the $550 billion disclosed.

factualhigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“It turned out the take rate was higher, and it turned out they were being modest about their TPV — it was much higher than 550, closer to one trillion.”

0.51

Corning benefits from scale-across data center wiring, which needs roughly 10 times more and thicker wire, and scale-up copper-to-fiber transitions within racks could be two to three times Corning's current opportunity.

forecasthigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“When that happens, that two- to three-x's Corning's opportunity.”

0.51

Each new Nvidia chip or rack uses 50 to 125 percent more power, which drives average selling prices and margins for power supply makers such as Delta and Advanced Energy for about four years.

forecasthigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“in the power supplies, every Nvidia chip or rack uses 50% to 125% more power. That drives the ASPs of Delta and Advanced Energy.”

0.51

Whale Rock bought Nvidia at about 4 times earnings in 2023, Tesla at about 5 times in 2019, and Apple at about 4 times, and got AWS effectively for free, because underappreciated earnings power can grow from $1 to $10.

factualhigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“When we were buying Nvidia in 2023, we were paying four times earnings. When we bought Tesla in 2019 for the car S-curve, we were paying five times earnings.”

0.51

PCB units are growing about 50 to 60 percent a year, layer counts are rising, and average selling prices and gross profits are rising too, with Elite Material providing the copper clad laminate that goes into these boards.

forecasthigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“so you've got a 50% to 60% CAGR just in the units. And the ASPs are rising.”

0.50

It is acceptable to miss the first one to three years of an S-curve, because if the top is a half-trillion-dollar market, growth can run for a long time, so investors do not need to be in at the very start.

normativehigh valuespeaker onlynovelty 2/4durability 3/4· Alex

“And by the way, it's okay to be late. It's okay to miss the first one, two, three years in a lot of cases”

0.50

For AI share metrics, the rate of change matters more than the absolute level, because a move from 10 to 30 percent with accelerating growth and margins is more valuable than a static high share, and AI models missed this when asked to plot the trend.

normativehigh valuespeaker onlynovelty 2/4durability 3/4· Alex

“So the rate of change is important. And that's incredible too, because if you go from 10% to 30%, and your growth rate accelerates and your margins accelerate.”

0.50

Whale Rock's research engine is a learning machine run by about 10 experienced people, averaging roughly 10 years of experience, that serves both public and private products, and its compounding of knowledge over 20 years is its core advantage.

factualhigh valuespeaker onlynovelty 2/4durability 3/4· Alex

“We call it the Whale Rock learning machine, and it's a group of 10 highly experienced individuals.”

0.49

Anthropic has maintained high market share in coding and built a strong enterprise brand, to the point that CIOs say Claude first, which Sacerdote cites as evidence of critical IP and durable advantage.

factualhigh valuespeaker onlynovelty 2/4durability 2/4· Alex

“Go talk to any CIO, and the first thing they'll say is Claude.”

0.49

Enterprise AI can command high prices because it replaces human labor, whereas consumer AI monetization is less clear, with advertising or a paid assistant being options, even though the consumer base is huge.

forecasthigh valuespeaker onlynovelty 2/4durability 2/4· Alex

“it looks like enterprise right now is much better, because you and I are willing to pay a lot, because it's replacing human beings.”

0.48

When a low-margin commodity supplier such as a contract manufacturer becomes an AI infrastructure partner, its visibility shifts from call-when-needed to multi-year roadmap design, taking it from roughly 5 percent growth to 35 to 50 percent top-line CAGR for four years with rising margins.

forecasthigh valuespeaker onlynovelty 3/4durability 2/4· Alex

“You've gone from a 5% grower, low-margin business to a 35%, 40%, 50% top-line CAGR for the next four years, with rising margins.”

0.47

Andy Grove argued that at strategic inflection points the data cannot be trusted, so investors must rely on intuition and anecdotal evidence, and visual, pattern-based observation is how the firm detects these shifts.

normativehigh valuespeaker onlynovelty 2/4durability 3/4· Alex

“Andy Grove says, when you have strategic inflection points, you can't trust the data. Strategic inflection points are about intuition, anecdotal evidence.”

0.47

Chip companies do not care which AI model company wins, so they benefit from the overall compute demand regardless of whether the frontier leaders or open-source players win.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Alex Sacerdote

“Chip companies don't care who wins, so that's another positive — and they'll benefit.”

0.45

AI demand has created shortages in DRAM, NAND, and PCBs, with supply roughly 30 percent short across these components, and high-bandwidth memory now requires stacking about 10 chips with roughly 10 times the input-output of prior memory.

factualhigh valuespeaker onlynovelty 2/4durability 1/4· Alex Sacerdote

“we're already 30% short all these things, as we are now.”

0.45

The AI infrastructure-layer S-curve is about 10 percent penetrated, and AI is still one of the best ways to play the theme, with the enterprise application AI market less than 1 percent penetrated.

factualhigh valuespeaker onlynovelty 2/4durability 1/4· Alex Sacerdote

“We have this infrastructure-layer S-curve, which we think is 10% penetrated.”

0.45

The bull case for B2B internet was right in direction but early by about 20 years, because the underlying infrastructure was not yet in place and it was realized later through SaaS.

factualestablishednovelty 2/4durability 3/4· Alex

“The underlying infrastructure wasn't in place for B2B to happen. It ultimately happened 20 years later, with SaaS.”

0.45

Oracle held a chokehold on the relational database market for decades despite free and open-source alternatives, because it had the database administrators and software tuned to its system.

factualestablishednovelty 2/4durability 3/4· Alex

“Oracle had all the database administrators. They had all the software that was tuned to work with them.”

0.40

AI's market could reach three to five trillion dollars, versus about $800 billion for cloud, which makes it the highest reward opportunity Whale Rock has studied, with higher risk too.

forecasthigh valuespeaker onlynovelty 2/4durability 1/4· Alex

“Cloud is maybe $800 billion; but this might be three to five trillion.”

0.39

Companies such as Google and Amazon grew without advertising, and Elon Musk has never advertised, because brand lowers the cost to acquire customers relative to their lifetime value, which is the core of the business model.

causalestablishednovelty 1/4durability 3/4· Alex

“You can also have brand, and brand is very important. Because Google, Amazon — they got to grow, and they never had to advertise.”

0.36

High-bandwidth memory, once a pure commodity, now stacks around 10 chips with input and output roughly 10 times higher, and Samsung needed years to do it, so suppliers must co-develop with Nvidia three or four generations in advance.

factualestablishednovelty 2/4durability 2/4· Alex

“Take memory, which used to be a pure commodity — this high-bandwidth memory is 10 chips stacked on top, and the input/outputs are 10x what they were before.”

0.35

Peter Lynch advised ignoring the historical price chart and focusing on the future, which Sacerdote uses to support the view that the past penetration of a curve should not limit one's estimate of its future growth.

normativeestablishednovelty 1/4durability 3/4· Alex

“Peter Lynch — I started at Fidelity, and he loved to mentor the kids, so I got some time with him. He said, white out the chart. It's all about the future.”

0.33

The private unicorn market is now bigger than most European stock markets, maybe even combined, and bigger than Germany's or the UK's, so investors must understand private companies that can be the largest in their space.

factualcontestednovelty 2/4durability 2/4· Alex

“the unicorn market is bigger than most stock markets in Europe — maybe even combined.”

0.33

AI agents may strengthen network-embedded systems such as Slack and systems of record such as Workday and CRM, because agents operate inside these platforms, though CRM going headless could reduce it to a database.

forecastspeaker onlynovelty 3/4durability 2/4· Alex

“maybe the next wave of AI will be these agents that use tools — they might operate inside the existing incumbent software tools”

0.31

Whale Rock's team identified the AppLovin opportunity before the market by following the company while private, knowing competitors and terminology, attending the Las Vegas app advertising conference and Cannes, and building a model supported by a strong relationship with CEO Adam Foroughi.

factualspeaker onlynovelty 2/4durability 2/4· Alex

“Michael and Sam really figured out the AppLovin story before anybody. They followed it when it was private.”

0.31

Whale Rock conducts 2,000 to 3,000 face-to-face management meetings a year, about 10 to 15 percent with private companies, and then focuses on a few companies it wants to learn deeply about.

factualspeaker onlynovelty 2/4durability 2/4· Alex

“So we do 2,000 to 3,000 face-to-face meetings with management teams a year, and about 10% or 15% of those are with privates.”

0.31

Whale Rock's analysts identified AppLovin before the market by following it while it was private, learning competitors and terminology, and building a model, a process Alex says AI cannot replicate.

factualspeaker onlynovelty 2/4durability 2/4· Alex

“Michael and Sam really figured out the AppLovin story before anybody. They followed it when it was private.”

0.31

Investors fail to hold high-growth winners like Nvidia because each good year triggers bubble fears and then a period of stalling, so the firm believes conviction built on the whole picture is needed to hold through the noise.

normativespeaker onlynovelty 2/4durability 2/4· Alex

“every time with Nvidia over the last four years, it's like, 'Oh, they had a great year — oh my God, it's got to be a bubble.'”

0.29

A major risk is negative public and some government sentiment toward AI, with only about 20 percent of people optimistic and Maine having banned data centers, creating potential for negative regulation even though the technology cannot be reversed.

factualcontestednovelty 1/4durability 1/4· Alex

“I think Maine just banned data centers, and only 20% of people are optimistic about AI”

0.28

Technology progress has come in nested mega-S-curves, including internet 1.0, mobile, cloud, e-commerce, and now AI, each building on the last, and Sacerdote believes AI is the biggest so far.

factualspeaker onlynovelty 2/4durability 2/4· Alex

“There are mega-S-curves and there are sub-S-curves. We've been lucky that we've had internet 1.0, mobile, cloud, e-commerce, and now AI”

0.28

At the 2013 Robin Hood Investors Conference, Whale Rock said AWS had already won because it had a seven-year lead, and that a competitor, Pepsi, existed but would not be big enough to displace it, which later came true as AWS reached about ten times the size of rivals.

factualspeaker onlynovelty 2/4durability 2/4· Alex

“we pitched Amazon for AWS in 2013 at the Robin Hood Investors Conference, and we said the bulls have no idea what they're sitting on — Amazon's won the war before it even started.”

0.27

Following Buffett and Munger's idea of compounding knowledge, Whale Rock's average team member has about 10 years of experience and two long-tenured members have been with Sacerdote for 18 years, so institutional knowledge compounds across cycles.

factualspeaker onlynovelty 1/4durability 3/4· Alex

“Munger and Buffett talk about compounding knowledge — we've been compounding that knowledge for 20 years.”

0.26

Sacerdote's father, an electrical engineer turned Goldman Sachs executive, joined Whale Rock as chairman after retiring from Goldman, and his modesty, warmth, and mentorship shaped the firm's culture until he died in 2011.

factualspeaker onlynovelty 0/4durability 4/4· Alex Sacerdote

“It's definitely my father.”

0.26

Owning Adyen required knowing Stripe thoroughly, because customers asked about Stripe whenever Adyen came up, so Whale Rock concluded the two were the Coke and Pepsi of cloud payments and had to own both.

causalspeaker onlynovelty 2/4durability 2/4· Alex

“But you can't invest in Adyen unless you know Stripe like the back of your hand.”

0.26

Whale Rock prefers to own private companies through their public listing and keep holding them, as it did with Nubank, instead of selling to the venture investors who typically exit once a company goes public.

normativespeaker onlynovelty 2/4durability 2/4· Alex

“The VCs are going to own, and then most of them are going to sell. We like to own it and keep owning it into the public market”

0.26

Whale Rock ran a focused long-short fund for its first 15 years, added a long-only product around 2015, offered an opt-in private allocation of 15 to 25 percent starting in 2015 and used it from 2020, introduced a hybrid fund up to 80 percent private in 2021, and recently launched a Mega Cap Tech Fund.

factualspeaker onlynovelty 2/4durability 2/4· Alex

“So the first 15 years, it was a long/short fund, and we wanted to be focused”

0.26

Whale Rock sold Apple in 2012 when about half of US adults had a smartphone, which was a mistake, because the big gains came in the zero-to-50 percent portion of the curve and Apple later compounded at about 20 percent through ancillary businesses and its App Store share.

factualspeaker onlynovelty 2/4durability 2/4· Alex

“And it was a mistake with Apple, because in the first five or six years of Apple, it was awesome.”

0.26

Amazon reached Walmart-size scale advantages in about five years, versus about 40 years for Walmart, because S-curves grow fast and digital businesses can scale quickly.

factualspeaker onlynovelty 2/4durability 2/4· Alex

“So they got a Walmart-size scale advantage in five years, versus 40 years for Walmart.”

0.25

Andrej Karpathy and Linus Torvalds, both strong coders, reversed their views: after the latest model, Karpathy went from writing 80 percent of code by hand to writing no code except in English.

factualspeaker onlynovelty 2/4durability 1/4· Alex

“Karpathy said last year's code tools could write 20%, and 80% would be handwritten. That flipped when the latest model came out, and now he hasn't written a line of code except in English.”

0.25

DRAM, NAND, and PCB supply are already about 30 percent short of demand, and if the L-curve is right, shortages will persist.

factualspeaker onlynovelty 2/4durability 1/4· Alex

“we're already short in the DRAM market, the NAND market, the PCB — we're already 30% short all these things”

0.24

The kindest thing anyone did for Alex was his father, a Goldman Sachs veteran of 41 years, who joined Whale Rock as chairman and oversight to help raise money and worked together with him for six years until 2011.

factualspeaker onlynovelty 1/4durability 2/4· Alex

“When I started Whale Rock — friends and family, he was the first call.”

0.21

Anthropic was a dark-horse startup that focused purely on enterprise, while OpenAI won the consumer market, and Google's Gemini could not be counted out.

factualspeaker onlynovelty 1/4durability 2/4· Alex

“Anthropic was this dark-horse candidate, this startup, and they focused purely on the enterprise. OpenAI had won the consumer.”

0.14

Whale Rock invested in Anthropic in August 2025 at around a 1 to 9 billion dollar valuation, before the company's revenue ramped from about 100 million to a billion and toward nine billion, and the firm stayed close to the CFO after investing.

factualspeaker onlynovelty 1/4durability 0/4· Alex

“So we made the investment at the 180 valuation, and I think they were hoping to get to a nine billion...”