Everything is now competing for the same dollar

Key takeaways

As recently as 2022, Germany and Japan could borrow for 30 years at close to zero. That world is gone, and AI is now bidding for what is left.

Over the past fortnight, I listened to founders, operators and allocators describe where their industries are heading. They covered four very different frontiers. Each described a capital requirement.

This matters for private markets, because all of those claims land on the same pool of money.

Operating leverage is moving from headcount to agent fleets.

On the 20VC podcast, Eight Sleep described engineers who each manage hundreds of coding agents. They described their email marketing as a team of zero.

On All-In, Uber and Snowflake were cited as evidence that companies can now grow without adding headcount. Baseten made a quieter point. Agents already choose and deploy models, while production changes stay gated by a human.

That detail matters. Across the new agent products, the rule that the agent drafts and a human hits send has become product positioning. Trust is now a selling point.

Some commentators say this will double GDP. Treat that as a horizon thesis. What the data shows today is narrower. AI investment is already a meaningful contributor to US GDP in Federal Reserve tracking, while aggregate productivity still lags the narrative. Mainstream estimates sit nearer 0.25 to 1 percentage point of annual productivity growth.

The direction is clear. The timing is not. Allocators should price both.

Gwynne Shotwell and Elon Musk told All-In that by revenue, SpaceX is now as much an AI business as a space business. Demand for compute rental is not dropping.

The roadmap they named is specific. V3 Starlink satellites arrive in 2026. AI compute satellites are targeted for 2027.

The logic is physical. In orbit, there is no permitting queue and no wait for generators. Solar power is constant, and cooling is simpler. SpaceX also paid $17 billion for EchoStar spectrum to connect satellites directly to mobile phones.

The AI trade is no longer just chips. It is launch, spectrum, watts and orbital real estate. Every layer needs funding.

On 8 September 2026, Stripe announced that Meta’s personal AI agent, Muse, can shop across the internet using Link, Stripe’s digital wallet.

The design is worth studying. Muse never sees the underlying payment details. At more than 1 million businesses that accept Link, checkout is instant. Everywhere else, Muse receives a single-use virtual card limited to the approved amount. The human approves every total in chat.

This is not two bots settling with each other. It is the first mainstream rail that treats an agent as an economic actor with a supervised budget.

Once agents can pay, distribution changes. Agents will increasingly be the ones your business sells to.

Martine Rothblatt of United Therapeutics told the Moonshots podcast that longevity will not come from a single miracle molecule. It will come from industrial organ manufacture.

She has four thymus programmes running in parallel. A pig thymus-kidney transplant is already on an FDA-cleared clinical path. Trials of a kidney built from the patient’s own cells are slated for 2028, with commercial-scale facilities planned in Minnesota and Texas. Rejuvenated organs are framed for the 2030s, and she says the timing may slip.

The allocation thesis is simple. Spending moves from treating sickness to keeping people well. The diligence filter for any bio platform is whether it has a clinical path and a way to scale to billions of people. A projected year for extending lifespan is not a filter.

Here is what connects all four. Each of these frontiers is capital hungry, and they are arriving as the cost of capital rises.

Goldman Sachs put a name to it on 17 September 2026. Its report, Competition for Capital, links AI and higher rates through one idea. They draw on the same pool.

Goldman does not call this a classic valuation bubble. US forward PE has fallen from about 22 times to about 19 times since early 2025. Earnings drove returns, not multiples. The bigger risk it names is an earnings boom that proves unsustainable, even while compute demand outstrips supply.

Its stance is neutral on equities over three months and overweight over twelve.

This is not a distant Wall Street story. Australian founders and fund managers raise from investors who price risk against the same global pool.

When the world’s largest companies are issuing at a record pace, a tide of cheap money will not carry anyone.

Capital is scarcer and more selective. Investors have more claims on their attention than at any point in my 17 years in this market.

In this cycle, the companies that communicate clearly and stay top of mind will recover capital that would otherwise be left on the table.

The deal does not get better. It just gets understood.

This article is general information only and does not constitute financial advice.