Field Note

Professional Identity and Market Divergence

How belief pools form in financial markets.

• Entity: InflectAI, Inc.
• Published: May 14, 2026
• Status: current

One year, I brought in $86 million against a $65 million quota.

It felt good. It felt like vindication. The number said I knew how to read the field, press the right deals, carry the quarter, and lead.

Then came the year when growth came in at 17% against a plan that needed 60%.

That number felt like an indictment.

There is no point pretending these numbers were only numbers. They were not. They were professional evidence. They told me something about the business, but they also told me something about myself. Or at least that is how they felt at the time.

That reflex did not begin in sales. It went further back, to the Army, where one of the first things drilled into you as an officer is that a leader is responsible for everything the unit does or fails to do. That doctrine has weight. It should. If the platoon fails, you do not get to point at the weather, the terrain, the supply clerk, or the private who misunderstood the order. You own the outcome.

That belief followed me into sales leadership. Forecast, pipeline, execution, coverage, deal risk, hiring, territory design, manager quality. Those became the instruments on the panel. If the number was strong, the panel said the machine was working. If the number was weak, the panel said something was wrong and the first suspect was usually me.

That discipline made me better at the job. It also taught me to mistake one kind of measurement for the whole state of the machine.

A professional role does not merely assign work. It trains perception. Over time, it teaches you what to measure, what to ignore, what counts as evidence, and which failures are yours to carry. The role gives you an instrument panel. If you stare at it long enough, you can begin to confuse the instrument panel with the world.

When I became a founder, how I saw the machine was totally different.

I still care about revenue. Of course I do. Revenue is oxygen with an invoice number attached. But the questions changed. I started thinking about capital, runway, product adoption, differentiation, customer friction, timing, and optionality. I asked whether something could be adopted easily enough to matter. Whether the customer pain was real or merely articulate. Whether the product was meaningfully different or just cosmetically improved. Whether the company had enough room to maneuver if the first plan was wrong.

Same person. Different role. Different salience map.

Belief Pools

That is the beginning of belief pools.

A belief pool is not just a group of people who happen to agree with each other. It is a professional population trained by its own instruments. The role teaches you what to count, what to ignore, what kind of error gets punished, and which question must be answered before anything else can matter.

A banker does not see like a short seller. A banker sees leverage, interest coverage, debt service capacity, collateral, maturity walls, syndication risk, and whether the borrower can survive the downside case long enough for the lender to be repaid. The banker's first question is not whether the story is beautiful. It is whether the structure can carry the debt.

A short seller sees something else. The short seller looks for contradiction: revenue quality that does not match the story, margins that depend on heroic assumptions, customer concentration, accounting stretch, insider selling, related-party weirdness, or a balance sheet that only works if capital markets remain friendly. The short seller's first question is where the story breaks.

A growth investor sees still another machine. The growth investor thinks in terms of total addressable market, penetration, revenue acceleration, gross margin trajectory, operating leverage, retention, and whether today's losses are buying a future monopoly position. The first question is whether the company can become much larger than the current numbers suggest.

A regulator has a different panel again. The regulator sees concentration, interconnectedness, consumer exposure, capital adequacy, liquidity, disclosure quality, operational resilience, and whether the failure of one actor can propagate into the system. The regulator's first question is not whether the trade works. It is what happens if too many people make the same trade at the same time.

These are not just different opinions. They are different instruments of perception. Each profession trains its members to see certain risks early and other risks late. Each profession has its own vocabulary for reality. Over time, those vocabularies become identity. People are paid, promoted, punished, and recognized through the instruments their pool uses to measure the world.

That is why belief pools are sticky. Changing the belief often means changing the instrument panel. Sometimes it means admitting that the thing your profession taught you to see first was not the thing that mattered most.

SoftBank As Test Object

SoftBank's recent AI financing is useful because it makes the belief-pool problem concrete.

On one side, SoftBank secured a $40 billion bridge facility to fund further investment in OpenAI and for general corporate purposes. That is not a small sign of market interest. A large banking group was willing to arrange financing around the company's AI ambitions.

On the other side, the cost of lending to SoftBank also carried a warning. Bloomberg reported that SoftBank sold $3.6 billion of junk bonds with a record 8.5% coupon on part of the deal, while other market coverage pointed to wider credit-default-swap levels and negative-rating-outlook pressure tied to the scale of its AI investments. The financing was real. The concern was real too.

Same company. Same AI story. Same broad fact pattern. Very different first questions.

The banker pool has an actual public trace. In the Federal Reserve's January 2026 Senior Loan Officer Opinion Survey, banks reported that they were more likely, on net, to approve commercial and industrial loans to firms benefiting from high AI exposure, and less likely to approve loans to firms adversely affected by high AI exposure. That is banker vision. AI exposure is translated into loan approval probability, borrower quality, financing need, credit structure, and repayment risk.

A banker looking at SoftBank does not have to believe the entire AI future in poetry. The banker can ask a narrower question. Can the facility be structured? Can it be syndicated? Is there asset coverage? Is the relationship worth defending? Is the maturity manageable? Can the downside case survive long enough for the lender to be repaid or refinanced? The banker's panel turns belief into terms.

The growth pool sees a different picture. For that pool, the relevant object is not only debt. It is convexity. If OpenAI, AI infrastructure, and the broader model ecosystem become one of the next platform layers of the economy, then SoftBank's exposure may be read as a costly but rational attempt to own a piece of the upside. The first question is not whether the financing looks heavy today. The first question is whether the future payoff is non-linear enough to justify the weight.

The short-seller or credit-skeptic pool looks for the opposite pattern. Michael Burry's recent AI-bubble posture is useful here not because he is automatically right. He is not scripture, despite the internet's best efforts to turn every famous skeptic into a minor prophet with a filing requirement. He is useful because the short-seller lens asks a different question first. Where is the contradiction? Where does the story depend on circular financing, rising marks, friendly capital markets, or future cash flows arriving exactly on schedule?

When credit protection gets more expensive, that pool does not see footnote noise. It sees the market charging more to insure against the story failing.

The regulator or systemic-risk pool has still another panel. The Bank of England's 2026 AI Consortium minutes discussed concentration risks from reliance on a small number of AI providers, models, and infrastructure; limited visibility over model design and update schedules; third-party dependencies; shared attack surfaces; and correlated failure modes. Federal Reserve Vice Chair for Supervision Michelle Bowman framed the supervisory problem in terms of material financial risk, third-party risk management, model-risk boundaries, and balancing AI's benefits against safety and soundness.

That pool is not asking whether SoftBank's trade is clever. It is asking what happens if the same AI infrastructure thesis becomes crowded across balance sheets, vendors, banks, data centers, power systems, and financial firms. It sees propagation before it sees alpha.

None of these pools has to be irrational. That is the point. The same object can look fundable, convex, fragile, and systemically concentrated at the same time because those words belong to different professional panels.

Divergence Before Convergence

The interesting question is not which pool is right at the first moment of divergence. The interesting question is how the fact moves.

Information does not propagate through markets evenly. It refracts. It passes through professional identity, incentive structure, time horizon, liability, and vocabulary. Some pools update early. Some resist because updating would invalidate the instrument panel they live by. Some translate the signal into their own native terms. Some wait until the market has already moved and then call the result obvious.

Eventually, belief has to reconcile with reality.

But not all at once.

That delay is where market structure lives. It is also where opportunity lives. If you can see how different belief pools process the same fact, and if you can measure where those pools diverge, you are no longer only watching sentiment. You are watching the structure of market perception.

This essay is only naming the first problem: how belief pools form and why they diverge. The next problem is harder. It asks when those pools collapse back onto reality, and why the collapse sometimes cascades.

Threshold dynamics and cascading collapse.

The sales number taught me a crude version of this before I had better language for it. The $86 million year and the 17% growth year were not just business outcomes. They were identity events because my professional pool had trained me to see through that panel.

The panel mattered. It still matters.

But the panel was not the world.