Capital Now Charges For Proof
The Decision Signal is a curated executive intelligence brief built to sharpen how decision makers see, think, and act. This week an $11 billion bond deal, a robotics listing slowdown and a $300 million issue that drew ten times the orders show the same thing: capital is available, and proof changes the terms.
Money did not disappear when rates moved higher. What changed is the price of getting it, and the price now varies far more from one borrower to the next than it did a year ago.
The Federal Reserve raised its target range by 25 basis points to 3.75% to 4.00% last week. The statement described economic activity as expanding solidly, domestic spending as resilient and capital investment as robust. Sixteen of eighteen policymakers expected at least one further increase by year-end.
I found it difficult to describe financial conditions as restrictive.
Kevin Warsh, Chairman, Federal Reserve, September 2026
If capital is still available at a price, the question is what the terms are saying about which parts of the plan the market is willing to fund.
- Cost: The market kept lending this week, but not on the same terms.
- Conviction: A commitment made months ago now has to be financed at today's price.
- Proof: Regulators and investors are asking harder questions about what sits underneath the valuation.
SoftBank launched over $11 billion of bonds to fund October's payment on an OpenAI commitment it made in February. The money was promised well before anyone knew what it would cost to carry.
The offering is $10 billion of dollar bonds across three tenors and €1 billion of euro bonds across two. Fitch rates the notes BB+. Proceeds fund the third and final $10 billion tranche of the $30 billion commitment announced in February, and cancel a bridge loan taken earlier to cover it. That sequence is the point. The investment decision was made seven months ago. What the market is setting this week is the cost of carrying it. SoftBank's dollar bond due 2031 now yields 8.2%, against 6.7% in January. The new notes are expected to price on Thursday.
↗ Bloomberg — SoftBank seeks over $11 billion in junk bonds for OpenAI betChina's securities regulator is steering humanoid-robot makers away from listing until they show recurring revenue and a path to narrowing losses. The instruction itself was never published.
It came through window guidance, the informal channel the regulator uses when it wants an outcome without writing a rule. At least six firms preparing to list are now waiting. The scrutiny followed Unitree Robotics, which rose more than fivefold on its Shanghai debut a month ago and has since fallen 55% from its peak. What is being questioned is the revenue underneath the valuations. Much of it comes from local-government-backed robot data-collection centers and joint ventures, where local authorities can fund 80% to 90% of the initial investment. The technology remains a national priority. The standard of proof changed.
↗ Reuters — China slows humanoid robot IPO rush as hype outruns realityBurford Capital raised $300 million in under six hours against an order book roughly ten times the size of the issue. The demand let the company cut the coupon well below where it started marketing.
The company is adding $100 million of its own cash and retiring a $400 million maturity due April 2028 in full, nineteen months ahead of schedule. Total debt falls by $100 million and the next maturity now sits more than three years out. Burford was explicit about the reasoning. Its own statement said credit spreads had tightened meaningfully but had not reached a level where a traditional long-term refinancing of the 2028 maturity was attractive. The demand rested on the record. Chief Executive Christopher Bogart pointed to seventeen years of producing high returns and nearly $4 billion of cash brought back to the balance sheet.
↗ Burford Capital — Burford Capital Update on Successful Capital RaiseCapital is separating conviction from evidence. Companies can still raise money and attract demand, and the market is now putting a clearer price on what has been proven and what still has to be believed.
SoftBank pays 8.2% to carry a commitment whose return is still ahead of it, against 6.7% on the same paper in January. At least six Chinese robot makers are held out of the market until the revenue behind the valuation is tested. Burford asked for $300 million against seventeen years of cash generation and was offered ten times that.
Where the evidence is thin, cost rises or access slows. Where the record is established, terms improve. Capital that followed the story now tests what supports it.
The wider the gap between belief and evidence, the more expensive that gap becomes.
The Decision Signal’s posture for the week of September 21, 2026 is Commit.
All five postures are live. Click on any zone to read this week’s signals from there, and what standing in it would cost.
◇ COMMIT: Separate what is already evidenced from what is still believed in every funded commitment. Decide before the next tranche releases which lines have earned more capital and which are waiting on proof.
Capital is still available to companies that can make the case. Higher rates changed the economics of borrowing without closing the market to anyone willing to pay for what they believe.
Access to capital is easily mistaken for validation. A company may be funded because investors believe in its future, because the return compensates them for the risk, or because a long record carries it. Those are three different reasons and only one of them is proof.
Most investment cases set out the expected return, the capital required and the execution milestones. Far fewer mark the line between what has already been demonstrated and what leadership still expects to become true. Which commitments are still funded on conviction, and what would prove them?
Getting the money does not prove the decision.
This week separated the two cleanly. SoftBank can raise $11 billion and still watch the cost of carrying the same commitment climb nearly a quarter since January. A Chinese robot maker can be a national priority and still be told to wait. Burford can be handed ten times what it asked for and say plainly that the money was not yet cheap enough to refinance.
Inside a company the distinction is harder to see. A project keeps its budget. An acquisition keeps moving. An AI program gets another round. Momentum can make the original case feel current long after the assumptions underneath it have changed.
Only a few commitments need this test. They are the ones still funded because leadership believes the case will eventually prove itself, and they are rarely the ones sitting on the risk register.
Go find them this week, and decide what evidence the next dollar requires before it is released.
Respectfully,
PJ Bickett
After reading this briefing, what is your immediate posture?