September 23: the cost of AI capital is rising

SoftBank launches $11B+ junk bond sale at record yields to fund AI bets

SoftBank kicked off marketing on one of the largest corporate junk bond sales ever — over $11 billion combined dollar and euro debt — offering record yields to investors to help fund its AI investments, including a follow-on stake in OpenAI.

Why it matters

SoftBank is one of the world's largest AI investors, and this deal comes on top of a $11.9 billion loan secured just over a week earlier and a $25 billion Arm-backed margin loan — a rapid stacking of debt across multiple instruments. Investors are demanding record yields to take the risk, which is a market signal, not just a SoftBank story: the cost of capital for AI infrastructure is rising even for the deepest-pocketed players. That cost eventually shows up in enterprise AI pricing, cloud contracts, and vendor budgets.

What this means for you

When an operator as large as SoftBank has to pay record rates to borrow for AI, it's a sign that "AI capex" is no longer cheap money — expect that cost to filter into pricing on AI tools and cloud services you use at work.

Finance: Watch high-yield spreads on AI-linked debt as a leading indicator — SoftBank's earlier bond sale this year already set a record 8.5% coupon, and rising demand for yield on new AI debt suggests investors are pricing in more risk around the AI buildout, not less.

Managers: If your company relies on AI infrastructure vendors backed by leveraged financing, budget for the possibility that financing costs get passed through in subscription or compute pricing over the next year.

Do this: Nothing to do yet — just be aware that AI infrastructure financing costs are climbing, and watch for pass-through effects in vendor pricing.

Signal 3/5· Pay attentionSources: SoftBank Starts Jumbo High-Yield Bond Sale to Finance AI Push, SoftBank Seeks Over $11 Billion in Junk Bonds for OpenAI Bet

Also worth knowing

Models & capabilities

Anthropic launches Claude Opus 5.5, cutting agent costs 40%

Anthropic released Claude Opus 5.5, its new leading model, which performs at roughly the level of Claude Fable 5.1 while costing 40% less to run than Opus 5.

Why it matters & what to do

Why it matters

For teams running long, agentic coding and knowledge-work sessions, price per token and tokens-per-task both drive the bill — Opus 5.5 cuts both. Cache reads, which dominate agentic coding costs, dropped 60%, making sustained multi-hour agent runs meaningfully cheaper to operate.

What this means for you

If your team runs AI agents for hours at a time, the same work should now cost meaningfully less without a quality trade-off.

Engineers: On an internal test rewriting HAProxy from C to Rust, Opus 5.5 finished faster and cost 51% less than Fable 5.1, and one tester completed a 680,000-line migration in under a day.

Finance: New pricing is $4/$20 per million input/output tokens and $0.20 per million cache reads, so budget forecasts for API-heavy workflows should come down accordingly.

Do this: If you're running long agentic coding or migration jobs on Claude, re-benchmark them on Opus 5.5 before your next budget cycle — the cost curve has shifted.

Source: Introducing Claude Opus 5.5

Signal 3/5· Pay attention

Policy & risk

Xi and Trump agree on one thing: don't slow down AI

Ahead of this week's White House summit, Bloomberg reports that Xi Jinping and Donald Trump — despite disputes over trade and Taiwan — are aligned on refusing to decelerate AI development, prioritizing global tech supremacy over safety warnings.

Why it matters & what to do

Why it matters

This is the clearest signal yet that great-power competition, not existential-risk rhetoric, is setting the pace of AI development worldwide — and it comes days after Anthropic's Dario Amodei publicly warned about the dangers of unchecked progress.

What this means for you

When the two governments with the most leverage over AI both choose speed over caution, expect regulation to stay light and model capability to keep accelerating faster than oversight can catch up.

Finance: Treat "AI slowdown" as a low-probability scenario in your planning — capital, chips, and policy are all still flowing toward acceleration, not restraint.

Managers: Don't wait for government-imposed guardrails to shape your AI adoption timeline; internal governance is now your team's responsibility, not a regulator's.

Do this: Nothing to do yet — just be aware that Washington and Beijing are unlikely to slow the pace of AI capability you'll need to plan around.

Source: Xi and Trump Seek Safe AI Without Slowing the Race for Supremacy

Signal 3/5· Pay attention

Money & markets

Drone maker Tekever hits $6.4 billion valuation as Europe's defense-tech boom accelerates

AI drone maker Tekever raised $580 million at a $6.4 billion valuation, becoming one of Europe's most valuable defense startups.

Why it matters & what to do

Why it matters

This isn't a niche defense deal — mainstream growth investors (a university endowment, a top-tier asset manager) are now writing nine-figure checks into autonomous weapons, and it's the latest in a run of huge European defense-tech rounds this year.

What this means for you

Capital that once avoided defense on ESG grounds is now flowing in at scale, and this shift is reshaping which sectors offer the fastest-growing jobs and equity upside in tech.

Finance: Defense tech is becoming an allocatable category alongside traditional growth equity, not just a specialist or government-only niche — expect more crossover funds and pension-linked vehicles to follow.

Managers: If you're hiring or retaining engineering talent, expect defense-tech startups to compete harder for the same AI and robotics talent pool as consumer and enterprise software firms.

Do this: Nothing to do yet — just be aware this sector is now a legitimate line item in growth-equity portfolios, not a fringe bet.

Source: AI drone maker Tekever hits $6.4 billion valuation

Signal 3/5· Pay attention

Money & markets

Micro1's data-labeling pivot hits $4 billion valuation

Micro1, a one-year-old AI training-data startup, has raised over $100 million at a $4 billion valuation — up eightfold in a year — with backing from frontier labs and two of xAI's cofounders.

Why it matters & what to do

Why it matters

A year ago Micro1 was a small AI recruiting business doing $7 million a year; its pivot into supplying human training data and "reinforcement learning gyms" to frontier labs shows where investors think the durable margin sits — not in building models, but in feeding them.

What this means for you

When capital chases the unglamorous layer — data, annotation, evaluation — faster than it chases new model releases, that's a signal the labs themselves see data as the binding constraint on progress.

Finance: Watch training-data suppliers (Micro1, Mercor, Scale AI, Surge) as a leading indicator of AI capex direction — their growth rates now rival the labs they serve.

Do this: Nothing to do yet — just be aware that "data provider" is now a legitimate, well-capitalized career and investment category, not a side hustle.

Source: This 25-Year-Old Raised Over $100 Million For His AI Data Startup At A $4 Billion Valuation

Signal 3/5· Pay attention

One line to sound smart

“When even the world's largest AI investors have to pay record yields to borrow, that cost eventually shows up in your vendor bills.”

Tool worth a look

Claude is the AI assistant this brief is built with — genuinely useful for drafting, summarizing dense material, and thinking through what a development actually means for you. An honest pick, not a paid link.

Try Claude →

Futureproof Daily is researched and written by AI against our editorial standards — see how we work. Sources are linked on each item. Nothing here is financial, investment, or legal advice.