⚡ From Chatbots to Agents: The 1000x Compute Shock Resets Macro, Tech, and Crypto

Key Takeaways

• Agentic AI is here: The shift from chatbots to autonomous agents has already increased compute needs by “a thousand times” and is broadening the beneficiary set beyond a single chipmaker.
• Macro regime check: Over the last year, the S&P 500 rose 15% while household net worth increased by $15 trillion. Near-term, higher inflation prints are expected; “you don’t want to be long the S&P 500 when CPI is that high” with a call that CPI will move above 4% YoY over the next two months.
• Productivity shock favors individuals: Solo operators can rent five top LLMs for $12,000/year and a couple of MacBook Pros for ~$5,000 to replicate teams — pressuring enterprise labor models and software seat-based pricing.
• DCF is breaking for software: Extreme uncertainty around AI-driven disruption undermines long-duration cash flow models; Bitcoin benefits as a growth asset untethered from DCF, while Ethereum remains more comfortable for traditional valuation frameworks.
• Labor market resilience: Despite AI-related job losses in select areas, structural shortages persist; it may only take 30–40,000 jobs per month to keep unemployment steady. The rate is up 80–90 bps from its low but remains near historic lows.
• Hardware, memory, and metals: The compute shortfall is acute; “DRAM prices are up four 500% from where they were in September.” Over the last six months, gold rose 20% and silver 60%; preference is expressed for Bitcoin, semiconductors, and silver over gold.

Background: Who’s Talking

Jordi Visser, Head of AI Macro Nexus Research at 22V Research, spent decades across derivatives and macro: Morgan Stanley in the early ’90s, equity derivatives and S&P options, an EM stint opening an office in São Paulo in ’97, then a decade running a macro fund and later a multi-strategy CIO. Today’s focus: how AI reshapes markets, growth, and policy.

From Chatbots to Agents: The Structural Break 🧠

“We have entered the Agentic side, which means the amount of compute we need has gone up by a thousand times already.”

• The pivot began in late November/December with Opus 4.5 and a rapid cadence of Anthropic releases. Capabilities now deliver workstreams measured in days instead of weeks, end-to-end.
• Agentic systems run continuously, autonomously, and at scale. The consequence: a permanent step-function higher in compute demand — and not just for one vendor.
• Infrastructure is the bottleneck: “We have no infrastructure for the compute… the supply can’t keep up.” Memory markets reflect this: “DRAM prices are up four 500% from where they were in September.”
• Early enterprise-grade stacks are emerging: “The first true enterprise data center… on Blackwell… ‘Lily Pod’… for Eli Liy.” The push spans frontier AI, biopharma, and eventually quantum.

Macro: Growth, Oil, and an Inflation Blip 📈

• Despite skepticism around AI, capital formation is real: S&P 500 +15% and $15T in household net worth gains over the past year anchor the bull case for continued AI investment.
• Oil’s macro bite is weaker than in the 1970s. Higher gas prices bring pain but “will not stop the AI spend.”
• Near term, inflation is set to re-accelerate: an “extremely high” import price print and a call that CPI will move above 4% YoY over the next two months.

  “You don’t want to be long the S&P 500 when CPI is that high.”

• Playbook: anticipate a positioning washout as recession chatter rises; historically, “when the pendulum shifts to recession is the best time to start buying.” The timing cited: within the next six weeks.

The Productivity Paradox: Solo Builders vs. Enterprises

• Productivity is most explosive outside large organizations. A lean tech stack — $12,000/year for five LLMs plus ~$5,000 for two MacBook Pros — can outperform legacy teams.
• Enterprise adoption implies labor displacement before margin expansion. Seat-based SaaS models tied to headcount growth are most exposed; expectations for ever-rising S&P 500 profit margins are challenged.
• Equity valuation stress:

  “By the end of this year, since you can no longer do a discounted cash flow on any software company… when AI agents… dominate workflows, you cannot value how these companies will be there.”

Why This Points to Bitcoin (and How Ethereum Fits) 🪙

• “Bitcoin is the endgame for growth assets.”

  As DCF reliability erodes for software, allocators seek growth assets untethered from cash flow models. Bitcoin captures flows from fiat-created wealth; Ethereum remains more compatible with DCF-style frameworks.
• Macro lens: velocity becomes central in an agent-led economy.

  “Velocity becomes the most important thing of GDP… velocity is what crypto is about.”

Labor Market: Shortages, Psychology, and the End of the Ladder

• Structural tightness persists: it may take just 30–40,000 jobs per month to keep unemployment steady. The jobless rate is up 80–90 bps from its trough but remains low.
• AI job losses have already occurred in select categories, but health care and services continue to absorb labor.

  “We’ve created zero jobs when you take out health care.”

• The acute issue is psychological, not cyclical:

  “There’s no more corporate ladder. AI is going to destroy that.”

  Upskilling on LLMs is the practical response:

  “Enterprises don’t have enough people that use the tools.”

Sovereign AI, Security, and Multiple Compression Risk 🛡️

“AI is a nuclear weapon… the most important part of the military.”

• Major powers will fund frontier AI regardless of cycles. Expect closer public–private entanglement over time — and the potential for multiple compression if control tightens.
• On the ad-based internet:

  “How are ads going to be worth anything when an agent is making the decision?”

  A bold horizon:

  “Human beings will not be making the majority of decisions within three years.”

• Long-term provocation: the arrival of humanoids alters firm structures and capital allocation.

  “Human beings are no longer the top of the food chain.”

Positioning the “Magnificent 7”: Hardware ≠ Software

• Do not treat the cohort as monolithic. There are three hardware names (Tesla, Apple, Nvidia) and four software/platforms (Google, Amazon, Meta, Microsoft). Views here skew constructive on hardware and skeptical on software.
• Nvidia remains central to the stack; the sideways price action reflects multiple compression despite outsized demand. Expect leadership on the next leg higher.
• On forward revenue signals:

  “Jensen… said we’ll have a trillion dollars of revenues… The market has their revenue at being closer to about 600 billion in a couple years.”

• Memory as a tell:

  “Micron… is trading now at a 4 PE off next year’s earnings.”

Metals and Commodities: Silver’s Outperformance

• Recent move: over the last six months, gold +20% and silver +60%. Preference is for silver over gold given its pervasive role in electronics, drones, and compute hardware.

  “It is effectively rare earth.”

• Framing: Beneficiaries of agentic compute expansion include semiconductors, memory, and industrial inputs such as silver.

Actionable Watchlist 🔎

• Inflation path: Monitor import prices and the next two CPI prints with the above-4% YoY call in view. Use any broad de-risking into recession chatter as opportunity.
• Compute capacity: Track data center buildouts, memory pricing, and supply-chain bottlenecks. The premise here: infinite demand for compute in an agentic world.
• Public–private AI dynamics: Watch for regulatory inflection that could compress tech multiples.
• Rotation lens: Distinguish hardware/commodities beneficiaries from software names facing DCF stress and seat-based revenue headwinds.

Tooling Up: How to Actually Get Agentic 🚀

• Voice-first: Build a daily conversational habit with LLMs on AirPods or in-car systems.

  “I have driven… for five hours with Grok… It is amazing to be in a car with the smartest person… because they know every subject.”

• Try agentic platforms: “Perplexity Computer is an agentic system… It will build dashboards. It will build websites.”
• Frontier models: Rotate among Claude, ChatGPT, Gemini and co.; capabilities are improving without higher subscription costs.
• Reps over theory: Replace passive content with hands-on projects; a short course (e.g., a three-hour intro) can unlock workflows that compound quickly.
• Make it social: Use LLMs for practical household planning (e.g., trip design) to lower the learning curve for everyone.

Quotes to Remember

“There are trillions of dollars… going to go through the economy over the next decade.”

“You’ve never had trillions of employees start at the job at the same time.”

“When the pendulum shifts to recession is the best time to start buying.”

“Human beings will not be making the majority of decisions within three years.”

Final Thought

The agentic transition is repricing cash flows, labor, and capital formation in real time. In a world where compute is the new oil and velocity is the growth multiplier, the market’s leadership, risk premia, and toolkits are being rebuilt from the ground up — with semis, silver, and crypto squarely in the flow.