Source: "Raoul Pal The Journey Man"

Compiled by: Felix, PANews

Macro investor and co-founder of Real Vision Raoul Pal invited Wall Street strategist Jordi Visser to join his podcast. In the interview, Raoul and Jordi analyzed how AI is creating a super cycle driven by computing, energy, agents, data centers, and intelligent explosive growth. They also discussed how cryptocurrencies, tokenization, and the data economy can open new markets.

PANews has compiled the highlights of the conversation.

Host (Raoul): Today, I have my regular guest and good friend Jordi Visser with me. This isn't an interview, but rather a joint exploration of what is happening and how to measure and seize opportunities. Jordi, how have you been lately? What have you been thinking about?

Jordi: I want to start with your conversation with Julian recently. You talked about the shift from "labor and capital" to "computing power and energy," which really resonated with me. In the past, developing a business required borrowing money, hiring people, and finding office spaces. But in a world of "computing power and energy," the rules are entirely different. I've been writing about the AI cycle, and in this new world, if we can't produce all the chips we need or obtain enough power, it will lead to bottlenecks and shortages in supply and demand. These bottlenecks may slow down the profitability of these companies, but not due to a lack of demand, rather because the demand is simply too great.

Host (Raoul): I've created an indicator dashboard to monitor the exponential growth of intelligent output per unit of energy. In the past, it was Moore's Law; now with the introduction of GPUs and AI, it is showing exponential growth on a logarithmic chart (double-exponential), which is what I refer to as Reed's Law (the square of the exponent). When you consider that the construction of data centers is only at 30% of what is expected, taking into account the competition between the U.S. and China, and that no single cutting-edge AI company can dominate, you will find that this will almost inevitably become a "super cycle." These bottlenecks will only slow its progress, and to break through the bottlenecks, we need to build infrastructure like power supply, which will represent the largest capital expenditure (Capex) cycle in human history.

Jordi: I completely agree. In surveys such as PMI (Purchasing Managers' Index) that examine business cycles, we often see that high PMI numbers are actually behind supply chain bottlenecks and rising prices, while new order indices may have already dropped to 50. But let me delve a bit deeper: how are we improving intelligence without having built enough data centers? This is because the algorithmic level, human feedback reinforcement learning, and reasoning capabilities have improved. It’s not a simple "drill more oil wells to get more oil" standard process; it's not "build more data centers to have more intelligence." This intelligence undergoes recursive learning and algorithmic improvement, which brings about dual-exponential growth. This speed of parabolic linear growth has become the norm, leaving many confused because this is rare in past emerging markets. From Jensen Huang talking in January at CES about the need to build an AI agent economy, to major players jumping in at the Morgan Stanley TMT conference in March, people began to realize how large the numbers could be. This phase is fundamentally different from the past three years of purely improving IQ; within the next year, the results of recursive self-improvement will shock you even more.

Host (Raoul): However, the market currently cannot keep all its attention and capital focused on a single stage; it hasn't yet grasped the significance of the AI agent economy. The total addressable market (TAM) of past businesses has always been humans, but now the TAM is infinitely expanding. So I think the market will see sector rotation; it won't always just be NVIDIA going up. When electricity and other bottlenecks arise, efforts will be made to clear all "roadblocks" to achieve intelligence per unit of energy, and capital and attention will also focus on those bottlenecks. For example, the peptide and genetic science technologies you've mentioned before will also see rotations.

Jordi: Exactly, electricity is an obvious bottleneck. This explains why NVIDIA and Siemens recently announced a collaboration in solid-state batteries in China, which require a lot of silver instead of lithium; everyone should pay attention to how much silver solid-state batteries need. If we had energy storage innovations to address peak electricity usage, the U.S. grid would actually be adequately powered through 2030. Huang has categorized AI into five layers: at the bottom are energy, chips, and infrastructure, then models, and at the top is the application layer. However, at the application layer, everyone is still focused on the old SaaS (Software as a Service) model. In the current application layer, the real absorber of most funds is actually "human software," meaning the capital attracted by Eli Lilly through GLP-1 weight loss drugs. They have data centers with thousands of GPUs on their campuses, and I believe that the cash flow generated by GLP-1 is actually financing the next phase of human biomedical software development.

Host (Raoul): It's like Elon Musk's used car business generating cash flow to subsidize new ventures. Speaking of solving bottlenecks and efficiency, Musk upgraded the voltage architecture of the Cybertruck from the traditional 12V to 48V to address the global copper shortage, directly reducing copper usage by 70%. When capital and attention are directed toward it, thanks to human and AI intelligence, we always find ways to circumvent obstacles. For instance, when facing oil bottlenecks, we invented shale oil extraction, and when encountering data center bottlenecks, we switched to fiber optics.

Jordi: People still haven't realized the significance of the AI agent world. Think about it; if someone in January had said that the Earth's population would suddenly increase by 7.5 billion people, we would think resources would be immediately depleted. But the AI agent world has brought billions, even hundreds of billions of new "thinkers" into the world, and they only consume one thing: computing power. Digital employees don't need to buy homes or send kids to college; people must change their old business cycle thinking. As Musk said, if there are billions of tireless agents solving problems for us, we will enter an age of abundance, where humans can choose whether they still need to work. These vast AI agents are running countless "Manhattan Projects," which will ultimately solve all problems.

Host (Raoul): And it's not just a single large model thinking. OpenAI has one billion users, and each user is using a different instance of this massive intelligence, which again adds to the exponential increase in intelligence.

Jordi: To assist myself, I've started building a personal "knowledge brain." I transcribe Huang's speeches and even upload the transcripts of hours of speeches by Eli Lilly's CEO David Ricks to integrate them. Utilizing the expertise of a crowd or individual as material for AI work is much more focused and profound than simply searching the entire interconnected world.

Host (Raoul): I am also building my "GMI brain" with a vector database, containing all the long-form content, video transcriptions, and my tweets I've written over the past 21 years. I’ve also created a tool called "Lens," which is based on my exponential age framework and first principles, capable of analyzing various issues from U.S. elections to the markets. But the problem I'm facing now is that I've done everything myself, and I simply don’t have enough time.

Jordi: I worked in a hedge fund for 20 years, and after it closed, I decided not to work for anyone else anymore and didn't want to manage anyone; asking investors for money and monitoring trades at midnight took the joy out of it for me. I realized that the global population of 8 billion doesn't know how to respond to the AI era, so I focused on content creation and helping people understand trends in accessible language. I have no employees, just one assistant, but the business is growing very rapidly and steadily. This is thanks to my use of AI agents handling everything. Building an AI business has astonishing profit margins and very low costs, allowing for rapid linear or even explosive growth.

Host (Raoul): So how are you squeezing out time for all of this? I feel like time has been completely consumed by learning new tools.

Jordi: I have more time than you because I don’t have to handle as many interviews, business trips, or manage company affairs. I also frequently ask ChatGPT to help me simplify things; I ask it, "Should I spend time on this new tool?" and it usually tells me, "Don't worry about it; it will be easier next month," which saves me a lot of time in trial and error.

Host (Raoul): That makes sense. I am currently focusing on building the underlying database engine, which I refer to as a "personal vault." This will store all your personal files, photos, and phone recordings, becoming your personal operating system that can be used for your brain or monetization in the future. Once every piece of information from computers and phones can be retrieved by AI in a second, many people still do not realize how disruptive this will be.

Jordi: I completely agree. A typical example is when a loved one passes away; handling inheritance documents as an executor can be a massive nightmare. In the past 18 months, I have been going through this, and I gathered all the documents into one folder linked to Claude Opus 4.5. When anyone calls asking about asset status or documents, Claude can immediately help me find the right answers and send them out, showcasing the power of future personal assistants and databases.

Host (Raoul): I am currently using a tool called Granola, which not only allows immediate meeting transcription but also connects with large models to become my long-lasting knowledge database. All conversations are fed into my knowledge brain, and it will never forget what we discussed last time. Right now, the biggest bottleneck in AI companies is "lack of lasting memory," and this long-lasting database layer breaks that limitation.

Jordi: I am currently running the Chinese model Kim K2.5 system on a Mac Mini and running GPT-5.5 on a top-spec M5 chip Apple laptop as my assistant and underpinnings of my portfolio technology algorithms. It can help me recall inspiration notes from two weeks ago while on a plane.

Host (Raoul): To avoid friction in data transfers between different devices, I just got the latest M5 Apple laptop, and now I take it everywhere.

Jordi: By the way, if you don’t hire anyone, the cost savings and extremely high profit margins of the AI business will prompt you to invest madly in hardware. To run the Hermes agent, I plan to buy an Nvidia DGX. After hearing Dennis Casabus's interview at Y Combinator, I am convinced that open-source models will become smaller and better, and ultimately we will move toward "edge computing" (Edge AI). In the future, running local models on your own devices and computers will be an extremely important part.

But learning to use AI (especially edge devices) is like learning to ski or golf; you must discard old habits and invest a lot of practice. When I encounter inspiration while walking and listening to podcasts, I immediately pause to record it using Whisper and then generate a draft of the article through ChatGPT. Therefore, people must buy the best phones and computers and treat them as an educational investment requiring money, using them constantly while walking, driving, or flying, to explore their workflows.

Host (Raoul): This constant thinking does lead to a lot of inspirations. I have been paying attention to the issues surrounding government debt and healthcare longevity. The net worth of American households is $180 trillion, while debts are only around $40 trillion, which is not too concerning. However, I am more concerned about how AI will profoundly impact the welfare system as life expectancy increases and healthcare spending takes up a rising proportion of welfare.

Jordi: The debt-to-GDP ratio will collapse. As for welfare and an aging population, while AI can indeed help lower welfare costs, the more difficult challenge lies in how to help the aging population regain their sense of value in the community or the economy. The boundaries of "work" have blurred; many YouTubers’ podcasts aim to capture human attention, which is a typical job in the post-AI era. Facing future work anxieties, I recommend reading "The Daily Stoic." Humans have worried for thousands of years about "what if we don't need to work," but humans always find ways to adapt and evolve.

Host (Raoul): I completely agree. Regarding debt resolution, I think of another key: tokenization. Two-thirds of the world's enormous assets (such as real estate, private equity, venture capital, art, etc.) are illiquid. Once these assets are tokenized, they will bring transparency and liquidity to dormant assets; once there is circulation speed, GDP will inevitably rise. Therefore, tokenization is a key focus of mine concerning longevity, ownership, and welfare issues.

Jordi: I have written a series of articles on the "invisible economy" (i.e., the AI agent economy). Cryptocurrency tokens are essentially machine-readable information packages. Google processed trillion-level tokens last year, and now it has reached tens of trillions. To train AGI into ASI (superintelligent AI), they need to absorb everything digitalizable, including university and scientific data. The largest market on Earth will no longer be the market of human assets, but the data market that AI craves. In the future, countless AI agents will engage in millisecond API call transactions, creating an enormous yet invisible trading market.

Host (Raoul): This is an excellent perspective. Because of this major trend, we need to reevaluate the concept of "bubble." A bubble is a combination of price and time. If the Magnificent Seven (Mag 7) soar to $20 trillion in a year, that's a bubble; but if they rise two to thirty times over 15 years, that's structural dividend. People always feel that if the rise is too fast, it's a bubble, but in reality, the profit growth of large companies is synchronizing with the soaring stock prices, and price-earnings ratios have even decreased. Indicators across various industries are showing hockey stick-like linear spikes.

Jordi: Old traders who understand too much are often more likely to miss good opportunities. They carry too many memories of things like the "internet bubble" or the Great Depression of 1929, leading them to misapply and compare the current market. It's like using Claude large models to write code; sometimes admitting that you know nothing, discarding extra context, can actually help the system work better.

Host (Raoul): I'm curious about your perspective. Currently, the stock earnings expectations based on AI hardware and infrastructure are very good, which is disadvantageous for narrative-based assets (such as Bitcoin and cryptocurrencies) because funds are being drawn away. Many people have not realized the explosive potential of the AI agent economy, and now those large funds controlling trillions of dollars are getting in, generating very stable profits. This will not only cause funds to rotate normally into longevity-themed drugs and other areas but also poses a test for the cryptocurrency market. When traditional stocks provide such enormous returns, what force will cause funds to rotate back into the crypto market?

Jordi: First of all, even giant stocks may stop rising due to insufficient liquidity and attention; we saw such rotations between 1995 and 2000 as well. As you said earlier, "infrastructure bottlenecks" will lead these AI giants to experience a performance plateau as supply fails to keep pace with demand. When funds overflow, decentralized identity (ID) and blockchain technology that meets the attributes of AI agent trading will once again show their absolute advantages. I am particularly focused on Layer1 protocols. While some believe AI will disrupt the existing SaaS software ecosystem, I think people are still embedding AI into existing billing and accounting software through APIs every day; software still holds its value.

Host (Raoul): The AI hardware world does indeed face limits on commodity output; for example, semiconductors require specific photolithography gases and petrochemical raw materials. Once production encounters bottlenecks, the scale of hardware infrastructure expansion may not meet everyone's expectations, resulting in a portion of the Capex valuation bubble being squeezed out. I believe that cryptocurrency is entering a "third wave" under two premises: first, AI physical infrastructure encounters bottlenecks that lead to capital fleeing to seek secure targets based on the digital world, which only require capacity without physical expansion (like AI agent application layers); second, tokenization makes traditional massive dormant assets liquid, allowing institutional funds to start questioning the chips they hold and seeking to enter the crypto world. We are currently in a bottleneck digestion period that requires great patience. What are your thoughts on recent large-scale IPOs from tech companies?

Jordi: Companies like Google financing at this time is because this is a competition for limited capital. Companies like OpenAI, Anthropic, or SpaceX cannot borrow from credit markets; they can only draw blood from the stock market. I believe that these three mega-IPO deals may signal a near-term peak in infrastructure capital expenditures (Capex) transactions but do not represent a peak for the entire market; funds will flow to software and other targets.

Host (Raoul): Yes. In my report, I mentioned that we have just experienced an "AI capital expenditure buffet," and everyone has eaten well; now we need a 3 to 6-month consolidation and digestion period to reset capital. As these underlying major trends continue to ferment, if stocks like NVIDIA stop rising, the probability and speed of funds flowing back into cryptocurrencies and other assets will dramatically increase.

Jordi: Cryptocurrencies indeed also need this digestion period. It's like when the gold ETF was launched; the approval of Bitcoin ETF and support from the U.S. president prematurely exhausted massive demand. Given that global liquidity only expands by 10% each year, the capital that has been front-loaded must inevitably take a year to digest. After the brutal bear market of 2022 and the emotional release of 2024, it is unavoidable to enter this painful consolidation period.

Host (Raoul): But when we attend large cryptocurrency conferences like Consensus in Miami, you find that there are almost no retail investors (although not many retail investors can afford the expensive tickets); the venue is filled with banks, financial institutions, and traditional forces focused on stablecoins and asset tokenization. This proves that the underlying technological trend is extremely strong, and once the infrastructure dividend of blockchain space begins to be realized and monopolized by a few core targets, the price of tokens will ultimately rise.

Jordi: Many people unfortunately get lost in the "trees" of short-term trading and overlook the entire "forest." If you've read "Elliott Wave Theory," you know we are waiting for that "third wave" that will make you a lot of money. Just like I previously caught Micron Technology's big surge, I am patiently waiting for the arrival of that "third wave" (also known as the Banana zone) for the crypto market.

Related reading: Conversation with Wall Street Strategist: AI-Driven Deflation Accelerates Capital Flow to Scarce Assets

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。