10 posts from 6 reports.
The Federal Reserve raised its benchmark rate by a quarter point to 3.75-4%, partly reversing the three cuts it made last year under pressure from President Trump. Chairman Warsh moved because inflation has picked back up. The Fed's preferred measure is running at 3.7%, nearly double its 2% target, driven by the jump in fuel prices from the Iran war, tariffs and the AI boom. The committee said the move would support a timely return of inflation to target, a goal it has now missed for five and a half years.
The signal for the months ahead matters more than the move itself. The Fed's own projections show 12 of 19 officials expect at least one more quarter-point rise before the end of the year, and four expect two. This is a cycle, and the market spent last year positioned for the opposite. The takeaway is to lean toward companies that benefit from higher rates or hold up against inflation, and to be careful with long-dated bonds. The risk to the call is a sharp weakening in jobs, which would give the Fed cover to stop.
Treat this as a real hiking cycle. Favour Value, floating income and quality cash generators over long bonds.
One of the world's top-performing sovereign wealth funds says it expects an equity pullback. That is a notable call because the fund posted 14% growth over the period while holding less US technology than the market. In other words, it did well without leaning on the most crowded trade, and it is now cautious even though it was never fully exposed to the froth.
When a large, long-term investor with a strong record turns defensive, it is worth listening. The message is not to sell everything. It is to trim the most stretched positions, keep some cash available for a better entry, and avoid paying peak prices for the handful of mega-cap tech names that now drive the index. If the pullback comes, the investors with dry powder will be the ones able to buy it. The risk is the familiar one: markets can stay expensive far longer than caution suggests.
Take some risk off the table. Keep cash ready and be selective on expensive mega-cap tech.
JPMorgan estimates AI could drive more than $430bn of incremental security software spending over three years. The reason it will come fast is that it is being forced, not chosen. As companies adopt AI, the number of ways they can be attacked climbs, and attackers now use AI to find and exploit weaknesses before a fix even exists. The time between a flaw appearing and being exploited has collapsed, in some cases turning negative. That makes automated, machine-speed defence a requirement, which JPMorgan expects to drive a steeper adoption curve than the shift to cloud or mobile.
The bank came out of earnings season more positive on Okta, Palo Alto Networks, CrowdStrike, Varonis and Zscaler in particular. Its security coverage is already up about 32% this year against the S&P 500. Recent prices give the anchor: CrowdStrike near $235, Palo Alto near $374, Okta near $186, Zscaler near $192 and Varonis near $47. JPMorgan is more cautious on Fortinet, where it sees the firewall cycle already priced in. The risk is that most of this year's share gains came from higher valuations, so the earnings have to follow as pipelines convert into 2027.
Own the best-positioned security platforms: Okta, Palo Alto, CrowdStrike, Varonis and Zscaler.
Over the weekend the heads of the top AI labs called for slower development because the technology is getting dangerous. For markets that sounds like a warning. For security vendors it is demand. That kind of public alarm raises the urgency for chief information officers and unlocks security spending from outside the normal IT budget. JPMorgan frames security as a partner to the AI build-out rather than a victim of it, and as critical infrastructure for it.
The money is flowing to the platforms that can pull many products together. Palo Alto added roughly 220 new platform deals with net revenue retention above 120% on that group, meaning existing customers keep spending more. CrowdStrike's flexible Falcon contracts reached $2.29bn of annual recurring revenue, up 101%. Palo Alto also cited seven-figure firewall orders from sovereign governments, new cloud providers and AI labs, with agentic network traffic up nine times in nine months. The build-out of AI capacity turns directly into demand for inspecting and protecting that traffic. The risk is heavy reliance on a few large deals that can slip between quarters.
Favour the platform consolidators winning share as security spend concentrates. Palo Alto and CrowdStrike lead.
Decades of offshoring hollowed out US industry. Manufacturing dropped from about 28% of GDP in the 1950s to roughly 9% today. The number of major defense contractors fell from 51 to five, large shipyards from 19 to eight, and the US share of global chip manufacturing from 37% to about 10%. The pandemic, war in Europe and the Middle East, and now the power and equipment demands of AI turned that dependence from a theoretical risk into a national-security problem. Rebuilding domestic capacity is one of the few things both US parties agree on.
The scale is large and slow-moving. Apollo estimates it would take about $2 trillion of extra investment to return the combined manufacturing and defense capital base to its 2000s share of the economy, and roughly $6.5 trillion to reach 1980s levels. Government incentives cover only part of that, so companies and private capital fund the rest. The priority areas are where a supply shock hurts most: energy, semiconductors, aerospace and defense, rare earth minerals, pharmaceuticals, and the robotics and automation that make domestic production affordable. This is a capital cycle that could shape where money flows for years. The risk is uneven progress, with delays from permitting, labour shortages and power constraints.
Own the reindustrialization capex cycle: energy, semiconductors, defense, rare earths, pharma and factory automation.
The rebuilding has clearly begun. US construction spending on manufacturing and energy projects tripled in three years to about $250 billion. The catch is how concentrated it is. Most of that increase went into computer and electronics plants, especially semiconductor sites and the infrastructure tied to the AI compute build-out. Investment across the wider manufacturing base has been far more modest so far.
That gap is the opportunity. If reindustrialization is real and policy-backed, the spending has to spread from chips into the supplier networks, the electrical and power equipment, the workforce and the surrounding industrial capacity needed to make domestic production work. Apollo frames this as part of a broader global industrial renaissance, driven by new supply chains, rising power demand and national-security priorities. For investors, the read is to look one step out from the obvious semiconductor names toward the companies that equip, power and supply the factories. The risk is that some of this capacity gets built ahead of real demand.
The next leg is the broadening beyond semiconductors: power equipment, suppliers and industrial automation.
Franklin's innovation team makes the bull case for AI plainly. Most technologies follow an S-curve: a burst of improvement, then a plateau. The lightbulb, the jet engine and the iPhone all did this. The semiconductor is the rare exception, doubling in power roughly every two years for six decades. Using an old chess parable, that puts chips into the "second half of the chessboard," where each doubling dwarfs everything before it. That is the territory of industrial revolutions.
Moore's Law is finally slowing as transistors hit the size of a few atoms and chips run too hot. Yet compute power is still accelerating, because the gains now come from whole systems rather than a single chip. Nvidia's AI compute has gone from 2,000 units of performance in the H100 (2022) to 9,000 in Blackwell (2024), with Rubin targeting 50,000 in 2026. The lesson of AI research is blunt: more compute wins. If that holds, spending on AI infrastructure has years to run, which cuts against the fear that the boom is about to plateau. The way to play it is to own the full stack of chips, memory, networking and power, not to bet on one component. The risk is that the payoff on this spending arrives slower than the capacity does.
The AI infrastructure runway is long. Own the whole compute system, not a single chip.
Healthcare has always been treated as a defensive sector with steady demand. Franklin argues that is changing into a growth story. The US spent about $5.3 trillion on healthcare in 2024, nearly four times the level in 2000, and it is heading toward more than a fifth of the whole economy by 2033. Demand keeps climbing as the population ages, yet the system cannot add doctors fast enough. Healthcare productivity has barely improved in decades while costs have soared, and administrative waste alone eats an estimated 15% to 25% of all spending.
That mix of huge, rising demand and deep inefficiency is what makes it investable. AI is the tool that attacks the waste: reading scans, automating paperwork and billing, speeding drug discovery, and letting a fixed number of clinicians treat more patients. The historical trade-off in healthcare was that you could improve cost, access or quality, but not all three. Franklin's case is that AI-enabled companies can now improve all three at once, which shifts revenue from paying per visit toward paying for outcomes. The read is to favour healthcare-technology businesses with recurring revenue and real adoption. The risk is regulation and slow procurement in a fragmented system.
Own the healthcare-tech names using AI to cut cost while improving care, from diagnostics to admin automation.
The recent selloff in Treasuries has an overlooked driver beyond war, inflation and debt worries. The buyer base has changed. For decades governments could count on pension funds and other buy-and-hold investors to absorb their debt. Those steady hands are pulling back for higher returns elsewhere, and faster-moving hedge funds have filled the gap. Hedge funds held about $2 trillion of Treasuries at the start of the year, double their level five years earlier, and a record 7% of the market.
Much of that is the basis trade, where funds exploit tiny price gaps between Treasury bonds and futures using heavy borrowing. That works smoothly until it does not. The New York Fed is studying whether these shifts have injected new risk into a market where $1.2 trillion changes hands daily. For investors the practical point is that Treasury yields can now move in sharp, fast jumps rather than gentle drifts, because leveraged traders can be forced to unwind quickly. Size bond positions for that. The tail risk is a disorderly unwind that briefly seizes up the world's most important market.
Expect more bond volatility. Size duration positions for sharp, fast moves rather than steady ones.
The Strait of Hormuz has stayed largely closed since the US-Iran conflict began, and the effect has now landed. Oil rose to $105.83, up more than $4 in a day, and American oil executives who spent months warning about a fuel crisis say it has arrived. This is the supply shock that strategists have been modelling, showing up in real prices rather than in scenarios. It is also feeding the inflation that just pushed the Fed to raise rates.
The investment read is to hold genuine energy exposure as both a return driver and a hedge. When the disruption is a physical shortage of oil, energy producers and the wider energy chain benefit directly, while the rest of the market wears the higher costs. The same shock is showing up elsewhere: US mortgage rates are pushing toward 7% again as bond yields climb, which threatens the housing recovery. The risk to owning energy here is a sudden de-escalation, since a reopening of the strait would pull prices back quickly.
Keep real energy exposure. The supply shock is no longer a forecast, and oil is above $105.
