The USS Abraham Lincoln left San Diego in November 2025 and did not come home. Its crew spent more than 250 days at sea, over 200 of them without a single port call, because the war that started in February kept extending a deployment that was supposed to end in June.

Last week the relief finally arrived. The USS George Washington came up through the Indian Ocean and took the watch in the Arabian Sea.

And where did it come from? The Pacific.

Here is what that means, and I want you to sit with it before we talk about money. For the first time since 2024, the United States has no aircraft carrier operating in the Western Pacific. Three of them are now clustered in and around the Middle East. The Theodore Roosevelt is tied up in San Diego. The Kennedy is doing sea trials at Newport News. The Carl Vinson has left San Diego and the Navy will not say where it is going.

So that is zero carriers west of Hawaii, in the one ocean where the actual long-term competition lives.

But this is not an accident, and it is not a scandal either. It is arithmetic. The United States built a fleet sized for one major contingency and told itself it could handle two, and February proved otherwise. When you send three carriers to one theater, you have taken them from somewhere else, and the somewhere else is always the quiet place.

Then today it got sharper, because the US cancelled its joint marine landing exercises with South Korea and gave force constraints from the Iran war as the reason. Not weather, and not some delicate diplomatic gesture toward Pyongyang. Constraints.

Most investors read a story like this as politics. So what is it really? An inventory problem — and inventory problems are the most reliably profitable thing in defense investing, because the government is the one buyer on earth who cannot walk away on price.

Think about what the Pentagon has actually learned in six months. It learned that a carrier can be held in theater for 250 days. It learned that Tomahawk stockpiles deplete faster than they replenish. It learned that presence in one ocean is absence in another. And it learned all of that in front of an audience in Beijing that has been taking notes since the first week.

So the spending follows the lesson. The FY2027 request came in at $1.45 trillion. Nuclear weapons and delivery systems are up 15% to $71.4 billion. Missile defense is up more than 25% to $85.8 billion, of which $17.9 billion is the Golden Dome fund alone. Last Monday the Department of War handed RTX $22.9 billion over seven years for one missile type.

Keep in mind what kind of money that is. It is not procurement of a new platform that might get cancelled in a future budget cycle. It is capacity — factories, tooling, second sources, and long-lead components for things that get consumed. And consumables have a wonderful property as a business: the customer always comes back.

If you have been treating the defense trade as a headline trade, tied to whichever conflict is on the front page this week, I urge you to change your frame. The conflict is not the driver. The gap is the driver, and the gap does not close for a decade.

And one more thing. Rheinmetall's order backlog stands at €80.5 billion, which is the largest in the company's history, and the shares are down about 26% over the past twelve months and sit 42% below their 52-week high. Read those two facts side by side and you have the whole opportunity in one company.

So the constraint is no longer the backlog. It is delivery.

If you did not claim a stake in SpaceX before the IPO…

Click here now and I’ll give you the name of a new startup that’s now backed by Elon Musk…

And that I believe will be the next hot IPO on Wall Street.

Bloomberg has just confirmed that…

This startup is moving towards “a blockbuster IPO.”

But you do NOT have to wait until the IPO.

Click here now and I’ll show you how to claim your pre-IPO stake for as little as $50.

Look, no other market event creates more overnight wealth…

Than an exciting tech company going public.

For example, early investors who got in at the pre-IPO stage had a chance to turn $1,000 into…

More than $1 million in Facebook when the company went public...

More than $1.6 million in Uber on its IPO day.

And more than $2.3 million in Google the day the company went public.

So click here now because I believe this IPO is imminent.

On Monday the Department of War awarded RTX's Raytheon unit a seven-year contract worth $22.9 billion to build Tomahawk cruise missiles. The goal is to take annual output from roughly 60 missiles a year to more than 1,000.

That is not an increase. That is a sixteen-fold expansion, and I would ask you to name one organization in history that expanded a production line sixteen-fold because things were going well.

Acting Secretary of the Navy Hung Cao framed it in the language of urgency: "Guaranteeing our naval and joint forces have the lethal firepower they need to deter aggression and protect the homeland is paramount. This landmark $22.9 billion investment will accelerate Tomahawk missile delivery to our warfighters at unprecedented speed." The work runs across Arizona, Alabama and Massachusetts, and it formalizes a framework agreement signed back in February — which tells you the Pentagon started worrying about this at the very beginning of the war, not at the end of it.

Asked about depleted stockpiles last Monday, the President was dismissive: "What we've used is peanuts; we have a lot of mid-level weapons."

But set that against the contract he had just signed. If the expenditure really were peanuts, would you restructure a production line for seven years to fix it?

And the Tomahawk award was not alone. On Thursday, US Transportation Command awarded FedEx and UPS $2.72 billion each, with Polar Air Cargo taking $272.5 million, for next-generation delivery services. On Friday it chartered $1.2 billion of roll-on/roll-off sealift across seven carriers, running from September 1 through August 2031. The Navy separately put up $997 million for petroleum storage and pipeline work.

Notice the pattern running through all three. Airlift, sealift, fuel. None of it is a weapon, and all of it is the ability to move weapons a very long way and then keep them supplied once they arrive.

In other words, the Pentagon is not buying for a war it expects to win quickly. It is buying for one it expects to sustain.

If you own the primes and nothing else, you are underexposed to the part of this that is actually scaling.

For a decade, Silicon Valley's relationship with the Pentagon was a moral argument conducted at company all-hands meetings. That argument is over, and you can see exactly when it ended by looking at where the money went last week.

Castelion, a hypersonic weapons startup, raised $800 million in equity plus $250 million in debt at a $13 billion valuation. The round was led by JPMorgan Chase, with Andreessen Horowitz and Carlyle participating. It was the single largest venture round anywhere in the world for the week ending August 21 — larger than any AI company, any biotech, anything.

Think about what it means that a commercial bank led a $13 billion round for a missile manufacturer, and that nobody in the market found it remarkable enough to argue about.

The rest of the week's list reads the same way underneath, with the satellite builder Muon Space taking $250 million from Eclipse Ventures and hardware crowding out software right across the table. And the broader market is running just as hot: 250 companies have reached unicorn status through August 15 of this year, against 193 in all of last year, with July alone minting 40 of them, the highest monthly count in four years.

But the defense-adjacent share of that is what changed. Venture capital spent fifteen years financing things that make attention cheaper. It is now financing things that make deterrence cheaper, because the government has become the one customer with an unlimited budget and a genuine emergency.

But here is the risk nobody in that market wants to discuss. Venture returns depend on exits, and exits in defense depend on a program of record, which is to say a line in a budget that Congress has actually passed. The FY2027 National Defense Authorization Act has not been enacted. The House passed its version 216 to 212 in July. The Senate failed to reach cloture 50 to 46, and there has been no floor action since.

So what exactly is underwriting a $13 billion valuation? Appropriations that do not yet exist.

That gap will probably close in the buyers' favor. But it is still a gap, and if you are buying into this theme through private vehicles at these marks, you should know exactly which line item your money is betting on.

Rheinmetall has a problem that most companies would kill for and no company can solve quickly.

Its order backlog reached €80.5 billion at the end of June. First-half revenue was €5.2 billion, up 39% from a year earlier, with an operating profit of €786 million. And it cut its full-year revenue guidance, to €13.7–14.2 billion.

How does a company with the biggest backlog in its history guide down?

Because Germany cancelled the F126 frigate programme outright on June 24, after roughly €2 billion had gone into it, and pivoted to as many as eight MEKO A-200 ships from TKMS instead. That took about €300 million straight out of the year. Meanwhile internal Bundeswehr documents reported last Friday show the Skyranger 30 air defense system slipping to mid-2027 with performance "not finally assessable," and a heavy weapon carrier running eleven months late on what the documents call insufficient maturity and quality.

And the shares tell the story better than any of it: around €1,155 against a 52-week high just above €2,000.

Now hold that against the money still arriving. France promulgated its updated military programming law on August 18 — Law No. 2026-791 — putting €449.3 billion on the table through 2030, an increase of €36 billion over the previous plan. Today the European Union approved €6.1 billion for Ukrainian defense procurement out of its €90 billion loan programme, aimed at air and missile defense, missiles, ammunition and radars, mostly from European suppliers. Norway pledged 85 billion kroner, around $9.1 billion, for next year.

Commission President Ursula von der Leyen: "As Russia intensifies its attacks, we are stepping up to help Ukraine protect its people and defend its skies."

And in Kyiv today, at a gathering of roughly 35 countries with no identified American representative, President Zelenskyy asked for at least 300 Patriot interceptors for the winter and put the arithmetic bluntly: "Over the entire of 2023, Ukraine received 675 missiles, and over all of 2025 – 364, even though Russian missile strikes have increased. For 2026, as currently projected, it is 264."

Strikes going up, and interceptors going down by more than 60% in three years.

So Europe's constraint has genuinely changed. It is no longer political will and it is no longer budget. It is machine tools, welders, explosive fill capacity and program management — and the companies that solve those get paid regardless of what any parliament decides next.

So buy the bottleneck, not the backlog.

Last Friday the Army put out a solicitation for something called IRON, the Intelligent Response and Orchestration Node, under a programme named Project Griffin. It wants AI agents that defend military networks on their own — seven functions in all, from blocking traffic at the firewall through patching vulnerabilities without a human in the loop. Solution briefs are due this Thursday, August 27, and up to seven companies advance.

Two concerns from the officials running it are worth more than the programme itself.

Wayne Sok, product manager for defensive cyber warfare, on the obvious hazard: "If we have a bunch of agents roaming around, and they're vulnerable, that just made it worse." Because an autonomous defender is also an autonomous attack surface, and the Army clearly knows it.

The second concern is stranger and more revealing. Officials flagged token costs as a scaling risk. Let me put that in plain English: the United States Army is worried that running AI models continuously across its networks will be too expensive per query. That is a sentence nobody could have written two years ago, and it points at something real, which is that inference cost has quietly become a defense planning variable.

Alongside it, the Defense Innovation Unit launched its Bridge Program last Thursday, opening shared classified facilities across nine states. Director Owen West said it will "open co-use classified facilities nationwide, slash cyber authorization timelines by half, and enable agile, end-to-end testing capabilities." And the bottleneck it targets is not technology at all. It is the accreditation paperwork that keeps perfectly good commercial software out of classified environments for years at a time.

And underneath all of it, the awards are getting less competitive. Palantir received up to $243.9 million in no-bid work through March 2027, with Deputy Defense Secretary Steve Feinberg directing components to identify further Palantir spending into 2028. The company's federal revenue hit $809 million in the second quarter, up 90% from a year ago. Across the whole government, no-bid awards were 14.8% of the total in the first half of this year, against 12.6% last year.

Most people see that as a story about one company's political access. But look again. What is it really? A procurement system that has decided it no longer has time to run competitions.

Speed has become more valuable to the buyer than price, and whoever is already inside the wire when that happens collects for years afterward.

On February 5 of this year, the last treaty limiting the strategic nuclear arsenals of the United States and Russia expired.

There was no ceremony, no replacement negotiated, no framework announced and no successor under discussion. New START simply ran out, and for the first time since 1972 there is no legal ceiling on the two arsenals that between them hold roughly 90% of the world's nuclear weapons.

Ask yourself when you last saw that discussed on a financial channel.

I want to give you the framework I think you need, because this is the kind of thing that gets ignored for years and then repriced in a single week.

Arms control did something that markets never gave it credit for. It made the future cheap to forecast. When both sides are capped and both sides can verify, a defense planner knows the size of the problem, and a treasury can budget against a known quantity. Remove the cap, and the planning assumption becomes the other side's potential rather than its actual, and potential is unbounded. So that is precisely how a defense budget becomes a defense spiral.

You can already watch the mechanism working.

The FY2027 request puts nuclear weapons and delivery systems at $71.4 billion, up 15% in a single year. Missile defense goes to $85.8 billion, up more than a quarter. Golden Dome, which was introduced with a $175 billion price tag, was reassessed by the Congressional Budget Office at $1.2 trillion over twenty years. Is that a cost overrun? No. That is what happens when the requirement gets defined by an adversary's capability instead of by a number written into a treaty.

The verification problem is spreading too. South Korea's defense minister said last Thursday that North Korea now holds 80 to 120 warheads. The figure the President has been using publicly is 57. When the range of credible estimates is a factor of two, every planner in the region has to budget to the top of it. The 2026 Non-Proliferation Treaty Review Conference ended in deadlock. On August 18, IAEA Director General Rafael Grossi said inspectors had found what he described as tons of nuclear material at an undeclared site in Syria, left behind by the previous government.

So what do you actually do with this?

First, understand that this is a decade-long spending floor rather than a cycle. Warhead life-extension programmes, the Sentinel missile, the B-21 and the Columbia-class submarine all carry fifteen to thirty year tails, and they are the hardest line items in any budget to cut, because cutting them requires admitting something out loud that no politician wants to say.

Second, look past the primes to the choke points — solid rocket motors, radiation-hardened electronics, specialty forgings, tritium handling. Each of those is a business with two or three qualified suppliers on earth, an enormous regulatory moat, and customers who have no ability to substitute. As Buffett has spent sixty years explaining, the question is never how exciting the product is — it is whether anyone else is allowed to make it.

Third, recognize what this does to sovereign balance sheets. Every government now facing this bill is facing it while its borrowing costs are the highest in a generation. Nuclear modernization in Britain already consumes roughly a fifth of the defense budget, and every pound of it is a pound not spent on the conventional forces that actually get used.

Fourth, and this is the one I would think hardest about, arms control was the last functioning piece of the US-Russia and US-China architecture. Trade agreements went first. Diplomatic channels thinned. Arms control survived all of it for fifty years because both sides understood it served them. It has now gone too, which means there is no remaining institution designed to slow an escalation down.

Every paid-up subscriber should be positioned for a world where the guardrails are gone and nobody has proposed new ones.

Horse, meet water.

Up to seven companies advance on the Army's autonomous cyber-defense programme. If you hold small-cap government software names, this is a live catalyst that almost nobody is watching.

It left San Diego and the Navy has not disclosed a destination. If it turns west toward the Pacific, the carrier gap was a scheduling artifact. If it turns toward CENTCOM, then four of America's deployable carriers are committed to one theater, and you should read every Pacific headline for the rest of the year differently.

Cloture failed 50 to 46 in July and nothing has moved since. Every defense valuation being struck in private markets right now assumes this passes. Watch the September calendar.

Berlin cancelled F126 after roughly €2 billion and pivoted to TKMS. If you want to know whether European rearmament can actually deliver hardware rather than announcements, this is the single cleanest test case on the continent.

The programme targets initial capability by mid-2028 against a CBO lifetime estimate of $1.2 trillion. If you are holding space and interceptor names, the schedule matters more than the topline — these are the contracts that get restructured first when dates slip.