Real Vision co-founder Raoul Pal hosted Wall Street strategist Jordi Visser on the Sept. 10 episode of Raoul Pal The Journey Man, where the two discussed AI, Bitcoin, tokenization, bonds, productivity and the dollar system. PANews compiled the main points from the interview.
They opened with bonds, not crypto
Visser said bonds were the first place to start because that has been at the center of his thinking. He argued that the market’s obsession with a supposed collapse in global 30-year U.S. Treasury yields has turned into what he called one of the biggest marketing-driven scares he has seen, amplified by social media.
When people point to the chart, he said, his focus is on total return. Long-duration ETFs holding bonds beyond 20 years are down only about 3% this year, in his telling, which he does not view as severe turbulence. The panic, he said, comes from the fact that yields are at their highest level in two decades, not from the actual total-return damage.
He said a recent discussion about massive AI capital spending led him to push back on worries over the bond market. Global PMI is rising, corporate profits are not just climbing but reaching unprecedented levels, and companies are doing that without adding workers. Hiring is close to flat, almost recession-like, while margins keep expanding.
Visser added that anyone who sold the Nikkei or other risk assets when 30-year yields topped out in May last year would have missed a 70% move. In his view, the yield panic associated with figures such as Ray Dalio has taken up too much attention. Higher yields can force out weak, overleveraged companies tied to the old economy, but AI will accelerate that process anyway, so he does not treat it as an outright negative.
Pal agreed. He said Treasury Secretary Bessent has already been clear that this is fundamentally a supply issue because hyperscale cloud firms and governments are issuing heavily. In his framing, the market needs a weaker dollar to draw foreign central banks into Treasuries. Yields will eventually be anchored, but what the system really needs is a steeper yield curve.
He said Kevin Warsh has to project a hawkish and credible stance, while Bessent keeps stressing that inflation will come in weaker than expected and productivity stronger than expected. Pal also pointed to DeMark indicators on weekly, daily and monthly time frames, saying they are stacked near levels often associated with extreme fear reversals. Any liquidation in the old, debt-heavy economy and in credit funds, he said, lies further out, at some later point when AI and yields intersect, not now.
AI is showing up in profit margins before it shows up in payrolls
Visser said Japan once engineered a steeper yield curve and saw banks begin lending again. He sees that as relevant now because this capex wave needs bank credit to scale. He noted that global bank stocks are up about 30% year over year. Markets discount the future, he said, and the rise in share prices, margins and earnings is already telling that story.
What stands out to him is the mix of conditions now in place: hiring has stalled, revenue is growing, fiscal deficits remain high, private investment is strong, and productivity is rising sharply.
Pal said the best evidence of a productivity jump is the combination of record profits and very weak employment growth. He referenced a piece he wrote using Amazon as an example of what a full AI-and-robotics economy could look like. Over the past eight years, he said, Amazon has turned itself into a productivity machine, with profits and gross margins moving sharply higher as those technologies were deployed.
Visser then went back to 2013, when he left traditional macro trading and went to Silicon Valley. He said he saw two developments colliding. China’s debt-driven growth model was hitting demographic and debt limits, while Amazon was trading at what looked to him like an impossibly high multiple. He went to Silicon Valley, visited Singularity University and came away with a different framework. Since then, he said, he has viewed the world as being in an era of time mismatch.
He used Japan as another example. When 30-year Japanese government bond yields made new highs, commentary was full of warnings that the world was ending and the yen carry trade was about to break. Eighteen months later, the Nikkei was still 70% above that point. The reason it matters, he said, is that one side of the equation is advancing at exponential speed, while the other side, including debt structures and GDP accounting, still moves at human linear speed.
He also said that ARR growth at Anthropic, OpenAI and Cursor is unlike anything seen before, while outside healthcare the global job market has recorded no net growth for 20 months.
The old nominal-GDP bond framework no longer explains everything
Pal said bonds were traditionally understood through nominal GDP. Under financial repression, the goal was to keep sovereign yields below nominal GDP growth.
Visser said historical data show that before 1997, the 10-year U.S. Treasury yield usually ran about 200 basis points above nominal GDP. With U.S. nominal GDP now growing at an annualized 6%, the old framework would imply yields around 8.5%, yet the market is closer to 4.75%. He traced that break to the 1997 Asian financial crisis, demographic change and the point when the internet became an investable asset.
He argued that many investors still miss a key point: for hyperscale technology companies, borrowing costs at 6%, 8% or 9% do not meaningfully damage business profitability. In fact, profits have kept climbing because these firms sit on large cash balances and earn more interest income when rates are high.
AI agents are multiplying fast
Visser said the world will see billions of AI agents over the next 12 months. Since he started using tools such as Grok, he said, he has built a large number of agents in just the past 10 days. The practical problem now is remembering what each one does. He said he already has 75 agents in circulation.
Pal said he even created a dedicated COO agent on Claude to manage all of his other agents because they keep breaking or disappearing, turning management into a full-time task of its own. He said he built six agents in three days on Grok.
In Pal’s view, people still do not grasp what happens once these tools are used deeply. The only real bottleneck left is that demand is unlimited. Last month alone, excluding his Mac subscription, he spent another $11,000 on AI compute and software, including tools such as Fable 5. Anyone expecting AI to create more free time has it backward, he said. It fills the schedule instead.
Tokenization, in their view, is becoming a national competition
Visser said tokenization now takes up most of his attention. He compared it to the race around AI, saying countries are beginning to compete over tokenization in the same way. Japan, he said, has already recognized that if it fails to move assets onto blockchain quickly, its capital markets will deteriorate further.
For him, market depth and democratic access to investment matter much more in a world that needs to raise more capital. Tokenization is, in his words, the ultimate answer to the financing problem around 10-year Treasuries and debt issuance: rather than limiting bonds to a narrow institutional base, they can be packaged so that all 8 billion people in the world can buy them directly.
Pal then asked what happens to the traditional asset-management chain, where corporate bonds and other products have long been wrapped into funds and sold to pensions or qualified investors.
Visser’s answer was blunt: adapt or die. He said Bitcoin has been the best-performing asset of the past 15 years, but traditional institutions have struggled with it because it is a black box and there is no human portfolio manager there to tell the story. Younger investors, he said, are far more comfortable interacting without a person in the loop.
He argued that tokenization makes it possible to fragment and recombine tens of thousands of uncorrelated return streams around the world into a portfolio generating 8% to 10% annualized returns, a very high Sharpe ratio and zero volatility, replicating the return architecture firms such as Citadel or Millennium spent heavily to build.
He also stressed the importance of continuous trading. This year, he said, events ranging from the situation in Iran to Hyperliquid to private SpaceX equity trading have all taken place in markets that run 24/7. Humans cannot compete with AI bots in a 24/7 market, he said, because people lose sleep and make emotional decisions.
Asset management may lose its middle layers
Pal extended that argument into asset-management structure. He described firms such as Millennium as motherships that allocate capital and control risk while countless pods underneath them generate return streams. In the future, he said, those pods may be individuals or AI agents, and the mothership will no longer need a huge back-office stack because everything will be tokenized.
That would collapse much of the intermediary cost and fee load now embedded in the industry.
Visser said all funds, individual return streams and AI agents will eventually move onchain and become investable. As turnover and the speed of capital circulation rise, pricing dynamics will shift. He pointed to Nvidia as an example. Since October last year, Nvidia has kept posting astonishing earnings, yet over the past 26 months its stock was mostly flat for 20 of them, with only four months of explosive upside.
His conclusion was that the old model of generating alpha by predicting earnings beats no longer works. In a market defined by faster turnover and heavier transaction flow, the consistent winners are the toll collectors, such as crypto infrastructure and trading venues, much like casinos make money by taking a cut of turnover.
Data, compute and energy are the next bottlenecks
Pal said nearly everything will turn into a market. AI agents consume energy and compute, but the heaviest demand may be for data. In that setup, a market for data driven by AI agents could become the largest market ever built, pulling in all available information.
Visser said absorbing all real-world video and sensor data requires massive compute and data-center capacity, and those resources are still in short supply. Political friction, physical bottlenecks and slower human adaptation may actually be useful, he said, because they slow the path to AGI and buy society time to think and adjust.
Pal brought up Tesla’s Full Self-Driving feature. He said he rented a Tesla in Napa, tapped the accelerator and let FSD take over, which showed him that the technology is already here. But humanoid robots delivering coffee on city streets are still years away because the compute needed for visual-spatial perception is not yet there.
Visser said mining and hazardous manufacturing will likely be the most important humanoid-robot use cases over the next five years. On that basis, if forced to choose between gold and Bitcoin, he would pick Bitcoin without hesitation. If humanoid robots enter mining at scale, he said, the extraction efficiency of gold, copper and oil would rise, costs would fall and supply would expand. Bitcoin, by contrast, remains a digitally scarce asset chosen by humans, with absolute scarcity that is not increased by automated mining.
What happens to the dollar if assets move onchain
Asked about energy bottlenecks around data centers and local political opposition in the United States, Visser said he is not especially worried. He pointed to an explosion of private-sector innovation in behind-the-meter power, including converting retired aircraft gas turbines into natural-gas generation units and placing data centers next to gas pipelines, bypassing the usual grid-approval process.
He also said tokenization is becoming a competition at the nation-state level. If a country fails to move onchain, its capital market will lose liquidity. Robinhood, he added, is a strong middle-layer example of the merger between crypto and traditional finance.
Pal then turned to the dollar. He said 75% of global investment capital in the MSCI index is already in the United States. If U.S. assets all move onchain, global retail investors could buy Nvidia or SpaceX with one tap from a mobile wallet. That would pull in even more global capital and make the dollar unusually strong in the short term. He said he agrees with Brent Johnson’s so-called dollar milkshake view that the dollar may die from its own strength.
Visser said the other side of that story is that stablecoins and blockchain rails will reduce U.S. control over the SWIFT system and over financial sanctions. Dollars would be distributed more broadly around the world. In his view, AI and crypto are the strongest decentralizing forces in human history, crossing borders and drawing younger generations deeper into the digital economy.
Both described AI as a live workflow tool, not a theory
Pal said the productivity shock from AI is no longer abstract. Earlier that morning, he said, AI scanned the GMI inbox and flagged two user suggestions. He liked them, used Whisper to dictate instructions, and had AI modify the code and push a new feature live on Real Vision in 40 minutes. Work that once would have required meetings, outsourcing and six to nine months was done inside a single cycle.
Visser said he has built his own knowledge brain using the YouTube API and X scraping. Every morning at 5 a.m., AI generates a custom briefing for him covering crypto and macro news.
He also described a problem from the prior week. On Wednesday, the software company he used for video recording shut down even though he had to record on Friday. He called what he described as an all-agent meeting inside Grokbot, similar to the executive meetings he used to run as a managing director at Morgan Stanley. The agents told him to switch to Google Slides and walked him through the process. He said he rebuilt the workflow in a few hours and produced a video presentation with 100 slides.
Pal said very few people around their age, close to 60, are using AI this deeply in traditional circles. Visser said the speed at which S&P 500 companies are disrupted and displaced will only rise in this era of extreme competition driven by AI.
On crypto, Visser said Bitcoin represents the world’s only truly scarce store-of-value asset. He added that over the next year, as tokenization and stablecoins expand, Ethereum may outperform Bitcoin as the base commodity layer of blockchain infrastructure.
Pal’s portfolio view was simpler. He said people spend too much time trying to pick entries and exits when they could just own Bitcoin, the Nasdaq, and perhaps some Ethereum, Solana, Zcash and gold, then stop staring at the screen. Since the Bitcoin white paper was published, he said, a portfolio split one-third S&P 500, one-third Nasdaq and one-third Bitcoin would have outperformed 99% of more complex strategies on an annualized compounded basis. Rather than obsessing over trades, people should focus on growing income from their main work and feeding that income into a simple compounding portfolio.
Visser closed by saying he has never sold any Bitcoin or digital assets to extract cash. He said he lives simply in Maine, drives a 2021 Tesla and does not chase luxury consumption. In his view, digital assets will eventually become at least a $100 trillion market, and Bitcoin will account for at least 33% of that total.