News vs. Noise: Rising Rates, Defense Growth and AI Security

Europe’s Defense Spending Is Showing Up in Company Results — but Spending Plans Still Have to Become Cash Flow

European governments have committed to substantially more defense spending, and some of that demand is now appearing in corporate results. Rheinmetall reported first-half 2026 sales of €5.23 billion, up 39%, along with a company-reported operating result of €786 million [1]. That gives us evidence that higher defense demand is moving beyond government announcements and into actual business activity.

But the company’s cash-flow results tell another part of the story. Rheinmetall reported negative €1.62 billion of operating free cash flow for the first half. The company pointed first to a shift in the timing of customers’ advance payments, higher inventories and higher receivables (money customers owe but haven’t yet paid), while it also kept investing to expand production capacity [1]. That distinction is important because a company can be growing rapidly and still consume cash while it buys inventory, builds capacity and waits to be paid [2].

There is also execution risk. Rheinmetall lowered its 2026 sales outlook by €300 million, to a range of €13.7 billion to €14.2 billion, following the cancellation of the F126 frigate program [1]. A single program was enough to change the forecast.

It is easy to hear that European defense spending is rising and conclude that every defense company should benefit automatically. The reality is more complicated. As advisors, we would want to distinguish among government spending commitments, actual contracts, company revenue, profits, and ultimately cash flow. Those figures are connected, but they are not interchangeable. The larger lesson applies well beyond defense: a strong industry trend can create opportunities without eliminating company-specific risks.

Long-Term Interest Rates Rose Again — and Homebuyers Are Feeling It

The Federal Reserve controls an important short-term interest rate [3], but it does not directly set the longer-term rates that influence mortgages and many other borrowing costs [4]. That distinction became especially relevant last week.

The 10-year Treasury yield rose from 4.96% on September 21 and 22 to 5.11% on September 23 and 5.18% on September 24 [5]. The 30-year Treasury yield also moved higher [6]. At the same time, Freddie Mac reported that the average 30-year fixed mortgage rate reached 7.03% on September 24, up from 6.95% a week earlier and 6.30% one year earlier [7].

Mortgage rates often move in the same general direction as the 10-year Treasury yield [8], which means changes in the bond market can translate fairly quickly into higher or lower borrowing costs for households. For someone buying a home, refinancing debt or evaluating a major purchase, even a relatively small change in interest rates can materially change the monthly payment.

For investors, rising long-term yields can also affect bond prices, stock valuations, and financing costs. Rather than focusing on one week’s move, we would watch whether inflation remains persistent enough to keep upward pressure on rates and whether economic growth remains strong enough to absorb those higher borrowing costs.

Google’s AI Security Test Exposed a Familiar Problem in a New Setting

A recent Google security test offered a useful reminder that some of the risks surrounding artificial intelligence are not entirely new. During a cybersecurity exercise conducted in May, Google’s Gemini AI agent was supposed to attack a fictional company. Instead, the system reached three real businesses because the fictional company’s name matched an actual business and the testing environment was not fully isolated from the public internet [9], [10].

One password was guessed, while two other sets of login credentials had been left in publicly accessible software repositories [10]. Google says Gemini stopped once it recognized that the companies were real, the businesses were notified, and no damage has been publicly reported [10]. The episode matters because it illustrates three separate security issues [11]: containment, because the test environment reached systems outside its intended boundaries; identity, because at least one password could be guessed; and credential management, because login information was left publicly accessible.

None of those problems began with artificial intelligence. What AI changes is the speed and scale at which software can search, test and act [11]. That is why concepts such as “zero trust” are receiving renewed attention. Under that framework, successfully logging in does not automatically give a user — or an AI agent — permission to access everything connected to a network [12].

Companies are also developing new products designed specifically to manage AI identities and permissions. Palo Alto Networks, for example, has incorporated CyberArk’s identity-security technology into a broader platform that includes protections for human users, machines and AI agents [13], [14].

The important investment lesson is not that every cybersecurity company will benefit equally from AI. The underlying need for better security appears very real. The harder question is which companies can turn that need into products customers are willing to pay for — and continue paying for. For clients, this is a good example of the difference between identifying a powerful long-term trend and assuming every company associated with that trend will necessarily be a successful investment.

AI Security Is a Real Need — but That Does Not Tell Us Which Companies Will Profit

The recent Gemini security incident reinforces a point we have discussed before: artificial intelligence is creating new cybersecurity challenges, and companies will need better ways to control what AI systems can access and what they are allowed to do. That represents a legitimate business need. What it does not tell us is which cybersecurity company will capture the economics of that need.

Microsoft is developing identity controls for AI agents [15]. Google offers its own security products [11]. Cisco has launched an AI Defense platform [16]. Specialist security vendors are pursuing the same need. Some protections may be sold as additional products, while others may eventually be bundled into software customers already use [11]. That difference matters considerably for investors because a technology can become essential without creating the same amount of new revenue for every company involved.

We saw a similar issue with the recent announcement involving Accenture and Anthropic. The companies said each expects to invest at least $1 billion over five years in their partnership [17]. That is a significant commitment, but it is not the same thing as saying Accenture has received a $2 billion revenue contract [11]. The announcement does not disclose exactly how much revenue, profit or cash flow Accenture will ultimately receive [17]. The sharp reaction in Accenture’s share price around the announcement [18], [19] is another reminder that short-term market moves can sometimes get ahead of what we actually know.

AI security and AI evaluation may become meaningful new industries. But when evaluating the investment implications, we want to separate four things: a real problem, a growing market, a company’s ability to win business, and the price investors are already paying for that expected growth. Those are related questions, but they are not the same question. That distinction can help investors participate in long-term themes without assuming every headline automatically strengthens an investment case..

A “Timber” Investment Is Not Automatically a Portfolio Hedge

Timber can sound especially appealing when much of the market’s attention is focused on technology and artificial intelligence. Trees continue to grow regardless of what happens to the next AI model, and timber investments may behave differently from technology stocks. But “different” does not mean “immune.”

One commonly cited timber benchmark, the NCREIF Timberland Index, is based on quarterly property appraisals rather than prices that trade continuously in public markets [20]. That can make the reported returns appear smoother because values are not being repriced every day [21]. Publicly traded timber companies behave differently. Their share prices can respond to lumber prices, housing activity, interest rates and credit conditions [22].

Historical data have shown relatively low correlation between companies such as Rayonier or Weyerhaeuser and Nvidia [22]. Correlation tells us how closely two investments have moved together historically [23], but a low correlation is not a guarantee that one investment will rise when another falls. Higher interest rates or a housing slowdown, for example, could affect timber companies at the same time other parts of a portfolio are under pressure [22].

There is also an important difference between investing directly in timber-related companies and owning a fund with “timber” in its name. The iShares Global Timber & Forestry ETF, for example, holds businesses across paper and forest products, packaging and specialized real estate investment trusts [24]. Those businesses do not all respond to the same economic forces.

Diversification is not about finding an investment that never declines. It is about combining investments that have different sources of return and different risks so the portfolio is not dependent on a single outcome. Timber may contribute to diversification, but it should not be treated as insurance against losses elsewhere in a portfolio. The more useful question is what an investment actually owns, what drives its cash flows, and how it has behaved during periods of real market stress.

The U.S.–China Trade Truce Was Extended — but the Harder Issues Remain

The latest meetings between the United States and China produced another extension of the countries’ trade truce, reportedly through January 10, 2027 [25]. That reduces the immediate risk of a new round of trade escalation, but it does not resolve many of the longer-running disagreements.

The meetings did not produce a detailed new tariff-reduction agreement, a comprehensive agreement on rare-earth supplies or a major change to U.S. restrictions on advanced semiconductor exports [26], [27]. In our view, postponing a hard decision can ease near-term worries without settling the underlying issue.

An agreement to keep negotiating can reduce near-term risk. It is not the same thing as a permanent agreement on tariffs, technology restrictions or trade rules. We would therefore look for concrete follow-through — written agreements, actual tariff changes, changes to export restrictions or evidence that announced purchase commitments are being carried out — before concluding that the long-term trade relationship has materially changed.