On November 16, the Federal Reserve Bank of San Francisco will convene what the wire copy is calling a "hearing" on artificial intelligence. Three words into that sentence, my antenna twitched. The Fed does not hold hearings. It holds conferences, roundtables, listening sessions — softer nouns that signal information gathering rather than adjudication. A "hearing" belongs to the vocabulary of the legislature and the courtroom. Either a journalist reached for a dramatic word, or something with teeth is actually scheduled in the heart of Silicon Valley. I have spent the better part of my career building frameworks around protocol trust and narrative velocity, and I have learned that the anomaly in the language is usually where the real signal hides. We don't just track trends; we hunt their origins.
Here is the thing about this dispatch: as a piece of reporting, it is nearly empty. Three facts, no agenda, no witness list, no official link. No year. That should make any analyst suspicious — and it should also make them pay attention, because the direction of the signal is unmistakable even if the details are thin. AI has stopped being a technology story. It is now a macroeconomic policy story, and the central bank of the world's reserve currency has decided it wants a seat at the table. Let me walk through why that matters more than the hearing itself.
The narrative cycle nobody priced
For most of the past three years, the AI narrative moved through familiar stages. First came the capability shock — models that could write, code, and reason well enough to unsettle knowledge workers. Then came the capital cycle: data centers, GPUs, power contracts, and a torrent of venture money chasing compute. Then came the public-market translation, where "AI exposure" became a valuation multiplier slapped onto everything from chipmakers to utilities.
What almost nobody modeled was stage four: the moment AI became an input into the Federal Reserve's reaction function. That is what a hearing — even a vague one — implies. The Fed does not convene on topics it considers peripheral. Its mandate is narrow and jealous: maximum employment and price stability. When a technology enters that frame, it is no longer a sector. It is a variable.
The wire copy hints at the framing with one phrase: "long-term supply shocks." Read that carefully. Central bankers draw a hard line between transitory disturbances and structural ones. A supply shock that is temporary gets shrugged off. A supply shock that is permanent forces a rewrite of the models — it changes estimates of potential output, of the neutral rate, of the entire glide path of policy. Calling AI a long-term supply shock is the Fed quietly admitting it may have to reprice the economy's speed limit.
Why I started paying attention to central banks
In 2024, after the Bitcoin ETF approval, I repositioned my own thesis work to bridge institutional capital and crypto-native narratives. I spent six months interviewing portfolio managers at Boston firms — the kind of people whose language is "yield-bearing collateral," never "community governance." I published a report I called "The Institutional Translation Layer," and the exercise taught me something uncomfortable: institutions do not adopt assets because they are elegant. They adopt them when the macro framework forces them to. Finding the human heartbeat inside the cold code is easy; finding the cold code inside the human heartbeat is the harder, more useful trick.
The same lens applies here. AI is being adopted by the Fed not because the Fed is excited, but because the arithmetic of its mandate leaves it no choice.
The two inflation paths — and the reverse logic
Here is the mechanism, and it is genuinely under-analyzed. AI touches inflation through two opposing channels.
On the cost side, automation compresses unit labor costs. A bank that replaces a research desk with a model, a hospital that triages with software — these push services inflation down. If this channel dominates, AI is disinflationary, and the Fed can afford to ease.
On the demand side, AI capital expenditure is enormous. Data centers, grid upgrades, power purchase agreements — this is investment demand on a scale that rivals a public-works program. If this channel dominates, AI is inflationary in the near term, and the Fed must hold rates higher for longer.
Which path wins is an empirical question the Fed cannot yet answer. But the market is already trading as if it knows. And here is the reverse logic that most AI bulls have not internalized: if the Fed concludes AI raises potential output, that raises the neutral rate, which raises the long-run discount rate, which compresses the present value of every long-duration growth asset. A productivity miracle, priced through the central bank, can be bearish for the very equities that celebrate it. That transmission chain is long, unglamorous, and almost entirely absent from the retail conversation.
The employment channel
There is a third path, and it is the one the Fed is legally bound to watch most closely. Its mandate names maximum employment alongside price stability. If AI displaces workers faster than it creates new roles — a structural, not cyclical, unemployment shock — the Fed faces an impossible trade: ease to protect jobs and risk inflation, or hold firm and preside over labor-market scarring it cannot repair with interest rates. Monetary policy has no tool for retraining a workforce. That is precisely why the framing matters: the Fed may be signaling that AI's employment effects are now a live input into its policy function, even though its instruments are blunt.
The measurement problem
There is a quieter complication the agenda almost certainly raises: the data itself. Productivity and employment statistics lag reality by quarters, sometimes years. The Fed would be steering by instruments that cannot yet see the thing it is trying to measure. A central bank that acts on a supply shock it cannot observe is a central bank trading on narrative — which, in its own sober way, makes it no different from the crypto market it regulates.
The governance land grab
There is a second story buried under the first, and it is about power rather than prices. The United States has no comprehensive federal AI law. Oversight is scattered across the FTC, NIST, the White House executive order, and a dozen congressional committees. Into that fragmentation steps the Fed — not with a statute, but with the softer, subtler authority of monetary policy and financial stability.
That is a land grab dressed as a listening session. By framing AI as an economic risk, the Fed positions itself to shape the terms of deployment without ever passing a rule. Call it soft regulation by framework. The San Francisco venue is not accidental; the hearing sits in the belly of the industry, which signals a desire to talk directly to builders rather than legislate about them from Washington.
And notice what the framing conspicuously omits. The classic AI safety lexicon — hallucination, alignment, jailbreaks — is absent. The Fed is not worried about whether a model lies. It is worried about whether a model destabilizes credit, employment, and price signals. Security is the canvas; liquidity is the paint — and the central bank is far more interested in the paint.
The blind spot worth naming
If I were advising the Fed, I would flag the thing the agenda seems to skip: financial stability. AI-driven trading, correlated model behavior, and concentration risk in a handful of providers are exactly the kind of systemic fragilities a central bank exists to monitor. When many institutions deploy models trained on similar data, reacting to similar signals, the herd becomes a single point of failure. I watched an algorithmic stablecoin unwind in 2022 because its "sustainable yield" narrative had no anchor beneath it; the death spiral was mechanical, not moral. AI in finance carries a similar shape — a narrative of efficiency layered over a fragility that only reveals itself under stress. The exit is easy; the narrative is the hard part.
The contrarian angle: this may not even be real
Now let me be honest about the evidence, because critical humility is not a stylistic flourish — it is a tool. The dispatch has no year, no author, and no official citation. The term "hearing" is inconsistent with the Fed's own vocabulary. And the source is a crypto-native outlet reporting on central bank AI policy — a cross-topic mismatch that is itself a signal, and not a reassuring one. It smells like aggregation, or worse.
So the contrarian reading is this: the specific event may be noise, but the trend it gestures at is real. Central banks globally are already studying AI's macroeconomic footprint. Even if this particular hearing is misreported or hallucinated, the underlying fact — that AI has entered the monetary policy frame — is not in dispute. The trap is treating a low-quality signal as high-quality intelligence. The opportunity is recognizing that the category of news is now permanent, even if this instance is not.
Where does that leave the capital? Watch the adjacent trade: compliance. If the Fed nudges AI deeper into the financial system's risk framework, the firms that build model-risk tooling and regulatory reporting become the quiet beneficiaries. It is an unglamorous corner, and unglamorous corners are where narrative velocity has not yet arrived.
Takeaway
The next twelve months will not be decided by which model tops a benchmark. They will be decided by how the institutions that set the price of money choose to interpret AI's productivity claim. Watch the Fed's language, not the hearing's headline. And ask the question nobody is asking yet: if the central bank decides AI is a permanent supply shock, what does that do to the discount rate baked into every "AI-native" token and equity you hold? The narrative is shifting from what AI can do to what the price of money will let it be worth.
