Video Summary
Enterprise AI penetration is still below 1% — adoption is poised for a steep, L-shaped surge.
Invest with S-curves: find companies on the right part of the adoption curve with durable moats and underappreciated earnings power.
Anthropic became a high-conviction private investment due to enterprise focus, model quality, and early strength in code generation.
Code generation is the first major commercial unlock for AI, creating sizable per-developer spend and a large addressable market.
AI demands are driving a hardware renaissance: specialized chips, high-bandwidth memory, liquid-cooled servers, and upgraded networking components are in shortage and high demand.
They observe enterprise AI penetration is extremely low (<1%) and recent breakthroughs (e.g., ChatGPT, code-generation tools) have produced a rapid, near-vertical adoption uptick — an L-shaped surge driven by immediate, high-value use cases.
Anthropic focused on the enterprise, demonstrated superior model quality and token economics, and showed early leadership in code-generation use cases that translated into meaningful per-user spend, differentiating it from many early startups and some large incumbents.
Coding tools quickly showed clear productivity and revenue potential (developers spending significant token dollars), creating a large, addressable market—estimates point to hundreds of billions from developer tooling alone—making monetization concrete and immediate.
AI can both augment and displace legacy software: some platforms may embed agents and become central, others risk being bypassed by headless, agent-driven integration. Ultimately incumbents must find AI-specific network effects or be at risk of commoditization.
Investors highlight a renaissance in specialized chips, high-bandwidth memory shortages, liquid-cooled AI servers, upgraded Ethernet/switching, and supply-chain tightness for DRAM, NAND, and PCBs — all creating durable demand and margin expansion for certain suppliers.
Many investors think linearly and focus on short-term quarters instead of long-term S-curve dynamics; they underestimate scaling laws, compute demand, and the speed at which generator use cases (like code) can drive revenue and infrastructure growth.
"When you get the right part of the S-curve, you get exponential unit growth."
The S-curve model demonstrates that with a strong business model, earnings can grow exponentially rather than linearly.
This concept reveals that while the world may not perceive exponential changes easily, understanding the S-curve can lead to accurate predictions about future growth.
The enterprise AI market is currently less than 1% penetrated, indicating significant room for expansion, which suggests we might be at the early stages of an L-curve growth pattern.
"When the gun went off with OpenAI and ChatGPT in November 2022, we took a massive deep dive with our team."
Following the launch of ChatGPT, a thorough analysis was conducted to identify new opportunities within the shifting technological landscape, particularly focusing on chips and infrastructure.
The analysts recognized that as foundational models evolve over the next few years, there might only be a few dominant players in AI, possibly creating an oligopoly similar to the cloud services market.
Over time, many early competitors in the AI startup space diminished, while major companies like Amazon and Meta struggled to maintain their footholds.
"We developed this thesis that it would be a three-horse race in AI."
The rise of code generation tools has shifted perceptions around AI's potential, particularly its capacity to replace manual coding efforts.
Tools like Microsoft C-Pilot indicate the growing demand and market for coding assistance, with significant expenditures by individual developers.
By analyzing market trends, the discussion suggests that there could be substantial financial opportunities to capitalize on within the coding sector, potentially amounting to a half a trillion dollar market.
"Anthropic has been able to stay ahead in coding."
Unlike traditional cloud computing services, which are often viewed as commodities, the discussion highlights the unique differentiators within AI models and their training methods.
For Anthropic, a focus on creating a comprehensive ecosystem around their API, including tools and software for enhanced usability, sets them apart in the industry.
The potential for AI technology is immense, but the development of unique competitive advantages will be crucial as the market evolves and matures.
"The enterprise AI market is less than 1% penetrated, but we're already seeing explosive growth."
The discussion reveals an urgent shift in how enterprises approach AI, suggesting that a significant change occurred this year, leading many companies to realize the necessity of AI implementation.
This is compared to the dawn of the internet, where businesses recognized the need for a website but faced challenges in developing them.
The potential for AI usage in enterprises is projected to rise sharply from 10 basis points to 15% over the next four years, indicating a strong trend toward widespread adoption.
The speaker notes that the current situation in AI resembles an L-curve, pointing to a steep upward trend in adoption rates.
"Mark Andreessen said there won't be enough compute available in the next four years."
A critical issue highlighted is the shortage of computing resources necessary for AI development, with companies like Anthropic struggling to meet their needs.
The rapid growth in AI capabilities is outpacing the available infrastructure, suggesting that industries must quickly adapt to the increasing demand for AI applications.
This scarcity of resources could significantly impact the next wave of AI innovations and applications in the enterprise context.
"We need to know these companies well because they can be the biggest players in the space."
The speaker shares insights into investing in private market companies, emphasizing the importance of building relationships and conducting thorough due diligence.
They recount an experience where understanding a company’s management and growth strategy enabled them to secure a substantial investment.
This approach included evaluating competitive market positions and leveraging extensive research to justify investment decisions.
The speaker mentions conducting hundreds of face-to-face meetings annually to assess potential private investments, underlining the role of rapport in obtaining desired allocations in these companies.
"S-curves are essential for understanding technology adoption life cycles."
The concept of S-curves is introduced as a framework for analyzing investment opportunities related to technology and innovation.
The speaker is invited to elaborate on how understanding the stages of market adoption—tinkerers, early adopters, and the early majority—can guide investment timing and decision-making.
There is an emphasis on the nuanced details of S-curves, suggesting they provide valuable insights into market dynamics that can significantly affect investor outcomes.
"When you have a strong business model, your earnings grow exponentially."
The correct positioning on the S-curve is vital for capturing exponential growth in earnings. In technology, strong business models can lead to significant advantages, and as earnings become increasingly underappreciated, they can spike from small numbers to substantial amounts, showcasing the potential for exponential growth.
Investing during phases of overlooked long-term earnings potential allows for acquiring excellent companies at low price-to-earnings ratios. For example, during purchases in 2023, Nvidia traded at four times earnings, and Tesla in 2019 was five times earnings.
The general market tends to underestimate future growth, often focusing on immediate quarters instead of looking deep into the future.
"Every technology follows the S-curve pattern; it takes time before the demand explodes."
The S-curve illustrates how technologies mature and gain market acceptance over time, with most major technologies taking years to reach their critical mass. For instance, smartphones existed for a decade before the iPhone's introduction changed consumer perceptions and drove demand massively.
Barriers to adoption, such as cost and usability issues, can stall technology growth until innovations like Apple's ecosystem streamline user experiences. Removing these barriers can result in a significant surge in demand.
"It's essential to understand the growth potential and when to exit a position after penetration peaks."
When assessing technology adoption, recognizing the slope of the S-curve helps in determining when to sell a position. Holding on too long past the peak can lead to missed profits, as growth rates typically diminish when penetration levels around 30-40%.
You must continually track market dynamics and signs forecasting when a curve may flatten, ensuring that investment strategies remain aligned with real-time growth expectations.
"Sometimes, you can miss the early years, and it's still worth investing later."
Observing anecdotal evidence and strategic inflection points is essential when evaluating when to invest in new technologies. Historical patterns resonate, as seen in mobile gaming and enterprise demands.
Engaging with industry events, like the Gartner IT Symposium, can reveal emerging technologies' growing demand before it becomes apparent in market data. This type of networking offers insights that could shape future investment decisions effectively.
Understanding that some markets evolve slowly while others accelerate rapidly helps tailor your investment strategy to each unique situation.
"The underlying infrastructure wasn't in place for B2B to happen — ultimately it happened 20 years later with SaaS."
The speaker reflects on their early experiences with investing, mentioning Amazon and the B2B internet boom, noting that infrastructure issues delayed B2B advancements.
They highlight that AI faces similar risks, especially with large, security-conscious companies that are often slow to adopt new technologies.
Cultural attitudes towards AI play a significant role; like the cloud, which initially faced skepticism over security, AI requires strong leadership and evangelism to encourage adoption.
"What's amazing about AI is you just open up the browser and it's there."
The rapid accessibility of AI tools sets it apart from previous technology trends, such as SaaS and cloud computing, which had slower growth rates.
There is an anticipated surge in AI usage, with projections for an increase in user engagement significantly from a small base. This creates what the speaker refers to as a "backwards L curve" indicating a steep growth trajectory in the near term.
"We look for the S-curve and try to find the one with a very powerful competitive advantage."
The speaker explains their investment strategy focusing on identifying companies that have separated themselves from competitors along the S-curve of growth.
They recognize that while many investors are wary of tech due to rapid changes and disruptions, certain intangible competitive advantages exist that can be even more robust in digital markets compared to traditional ones.
Examples of powerful competitive advantages include network effects, the ability to become an industry standard, and critical intellectual property.
"AI is by far the most complex and the fastest changing."
The speaker addresses the unpredictable nature of the AI market but stresses that the potential rewards can be immense, with market estimates in the trillions.
Companies like Anthropic and OpenAI demonstrate strong market shares and are innovating rapidly, showcasing the importance of maintaining competitive advantages, especially through intellectual property, branding, and advanced code capabilities.
The conversation reflects confidence in AI firms like Anthropic making strides in enterprise markets while building a strong reputation among decision-makers.
"On the internet, the leader goes bigger, faster, and wins."
The speaker asserts that companies which establish themselves as leaders tend to continue growing exponentially due to their scale and resources.
They emphasize that infrastructure, such as the computational capacity needed for AI, creates barriers to entry for many competitors.
Significant investment in research and development is paramount, as it allows leading firms to maintain their market dominance amid rapidly evolving technologies.
"When I look through your portfolio, I don't see a ton of big software companies; it’s hard to have the experience of building really useful tools and not think that maybe I could build a replacement for my company."
The conversation highlights a shift in perspective regarding software investment, particularly in AI applications.
Historically, the portfolio may have included a significant amount of software, but recently, a pivot away from those investments has occurred due to the unsatisfactory performance of AI products in the market.
There is a growing realization that many established software companies are struggling to produce AI applications that drive substantial revenue.
The emphasis on differentiating within a market dominated by larger firms suggests a saturation, where even innovative startups struggle to stand out.
"The old way of software is like using pen and paper; the new way is revolutionary, like a jet engine or the transporter from Star Trek."
Offering a perspective on the transformative potential of AI in software development, the speaker likens traditional software to outdated technologies compared to the advancements made possible through AI.
Enterprises are showing hesitance in prioritizing traditional software tools due to tighter budgets and quicker returns on investment with newer technologies, like language models and AI APIs.
As organizations opt for rapid deployment and immediate ROI, many software firms find themselves at a disadvantage, struggling to demonstrate value or charge effectively for evolving AI-driven products.
Despite the challenges, there are opportunities where AI integrations could solidify the position of existing software platforms.
"Even if the agents are running alongside established systems, they may become crucial for the organization's ecosystem."
There's an evolving landscape where AI could enhance the functionality of existing software tools, potentially making platforms like Slack integral to daily operations.
The discussion touches on whether companies might move toward headless systems, where AI directly interfaces with underlying data without a traditional user interface, potentially bypassing standard customer interactions.
The risk remains that while some existing tools are solidified by AI integration, they must continually innovate to avoid becoming obsolete.
Identifying the network effects specific to software platforms becomes crucial for determining which companies are best positioned to thrive amidst this transformation.
"For the past 40 years, nothing has changed in the data center, but now workloads are growing tenfold every year, pushing every hardware aspect to its limits."
The speaker emphasizes a significant shift in the hardware market, as AI demands a drastic increase in computational power and decommoditization.
After years of stagnant hardware innovation, the emergence of AI workloads necessitates rapid advancements and experimentation in hardware design and functionality.
As the technology evolves, there is potential for new market leaders to emerge due to increased competition in hardware optimized for AI, contrasted with the legacy systems that dominate current data centers.
Understanding how the industry adapts to and capitalizes on these changes will be critical for future investments in technology.
"We're in this renaissance of chips, requiring tremendous innovation at every aspect of the server."
The discussion highlights the ongoing innovation in the chip industry, emphasizing the need for high bandwidth memory and the significant advancements in input/output performance.
Companies like Celestica, which once faced challenges in a commodity-driven market, have now found success by maintaining specialized talent in supercomputing and becoming key suppliers for AI infrastructure.
Celestica's expertise is illustrated by its role as the sole supplier for Google's TPU servers, showcasing how they have capitalized on their historical strengths to adapt to modern demands.
"These AI server computers require liquid cooling and are running much hotter than before."
AI infrastructure demands sophisticated designs, as the shift from older servers to AI servers entails a substantial increase in both cost and complexity, with AI machines priced around $200,000 compared to older models at $5,000.
The critical reliance on components like high-performance Ethernet switches has created a booming market, as companies now require frequent upgrades rather than the past seven-year cycle.
The software layer in networking, particularly the open-source sonic layer from Celestica, demonstrates how integration and partnerships with companies like Broadcom lead to competitive advantages.
"We've gone from a 5% growth or low margin to a 35% to 50% topline CAGR for the next four years."
The market dynamics reveal a significant increase in demand across multiple sectors tied to AI, resulting in rising gross profit margins and enhanced visibility into future revenues from clients needing long-term partnerships.
The component supply chain is experiencing shortages, ensuring continued growth as companies like Corning, which produces fiber optics, adapt to fit the modern requirements of networking and AI infrastructure.
The need for advanced Ethernet connection solutions and thicker wiring plays a pivotal role in the scalability and reliability of AI systems, paving the way for remarkable opportunities in increasing ASPs and market share.
"We are already 30% short on the DRAM market, the NAND market, and the PCB market."
As AI technologies scale, the complexities of compliance, security, and operational needs compound, stressing the necessity for advanced solutions in asset management and investment sectors.
This environment exposes gaps and opportunities for investment, as the market is currently underestimating sectors vital to AI infrastructure, including memory technologies and related components.
Overall, there is a consensus that the demand for AI-related infrastructure will continue to grow, and investors who recognize these emerging trends are positioned to benefit in the long term.
"If you can see the whole picture and understand how these things are unfolding, it helps to have the big picture."
The discussion reveals a bullish stance on the impact of AI, emphasizing the importance of understanding the foundational models behind AI advancements. Analysts who lack a comprehensive view may miss significant developments.
Concerns remain about public negativity towards AI and potential government regulations, with statistics showing that only 20% of people are optimistic about the technology.
Despite these concerns, the speaker believes that much AI adoption will occur even if foundational models stagnate in their improvements.
"If AI sort of slows down in its improvements... it might be a race to the bottom, and it might not be good for the stocks."
There is a risk that if major AI players falter, it could lead to an oversupply of computational resources that might not be utilized effectively.
The importance of open-source models is highlighted, noting if established players like OpenAI hit a wall, the competitive landscape might change significantly, affecting financial projections.
"Historically, applications ended up being most of the market cap... but we think the application layer always comes later."
The focus is on the foundational layer of AI rather than the application layer, which is still seen as risky and unclear in terms of market development.
While companies such as OpenAI have released applications, there are lingering uncertainties about whether these applications can attain a sustainable moat in the marketplace.
Examples of up-and-coming companies, such as Sierra led by Brett Taylor, position themselves as potential successes, but the timing for widespread application adoption remains uncertain.
"We're meeting with as many companies as humanly possible... AI can be a great reporter, but it can't quite pick into the future."
The nature of research and analysis is evolving, with AI tools assisting rather than replacing analysts' roles.
Human relationships and in-depth knowledge about companies and industries remain irreplaceable, as effective analysts actively engage with management teams and rivals.
While AI systems help streamline note-taking and reporting, the unique insights derived from human experience and interaction continue to be pivotal in shaping investment decisions.
"When I like something, and then my analyst likes it, and then somebody who I really respect also likes it, that three legs of the stool can really help the conviction."
"For the first 15 years, it was a long short fund. We grew that to the scale that we wanted to."
"In 2015, we formalized the option for private investments, allowing investors to opt in or opt out."
"We just think there's a huge structural underweight of the largest tech companies in the world."
"We call it the Whale Rock learning machine, and it's a group of 10 highly experienced individuals."
"The kindest thing anyone's ever done for me is my father, who was an amazing mentor to so many people."
“Every investment firm is unique, and generic AI doesn't understand your process. Rogo does.”
Rogo is a specialized AI platform designed specifically for Wall Street investment firms.
Unlike generic AI solutions, Rogo connects to a firm's specific data and understands their unique processes.
This tailored approach allows Rogo to produce meaningful outputs that can enhance operational efficiency and decision-making for investment firms.
“The best AI and software companies from OpenAI to Cursor to Perplexity use WorkOS to become enterprise-ready overnight.”
WorkOS is a platform that enables companies to expedite their journey to becoming enterprise-ready by providing necessary infrastructure.
By using WorkOS, businesses can bypass unglamorous foundational work and concentrate their efforts on developing their products.
This streamlining can significantly reduce the time it takes for technology companies to scale and improve their competitiveness in the market.
