Bottleneck #3 Rolls Royce $RR.L Most people still think Rolls Royce makes cars, It doesn’t, it sold that off years ago. The Rolls Royce on the stock market makes engines for big long haul planes, the nuclear reactors inside Royal Navy submarines, and a load of generators. And it’s absolutely flying right now, with £2.5bn of profit in the first half of this year and up 46%. But the bit I want to talk about barely makes any money yet. THE BOTTLENECK Rolls Royce has spent decades building reactors for submarines, and now it’s using that to build small modular reactors, also known as SMRs. Each one makes 470MW, which is enough to power a whole data centre campus by itself, and about 90% of it gets built in a factory instead of on site. That matters because normal nuclear plants are a nightmare to build. Look at Hinkley Point C, years late and billions over budget. The idea here is to make nuclear more like building jet engines, the same thing over and over. And it keeps winning. The UK picked it over GE Hitachi and Holtec to build Britain’s first SMRs, three of them at Wylfa in Wales. Sweden picked it this summer, the Czech Republic wants up to six, and Rolls Royce says it’s won every competitive nuclear tender it’s entered in Europe. THE BULL CASE AI data centres need power that never switches off, and that’s exactly what nuclear does. But what I like most is that the rest of the business is paying for all of this. It’s expecting £4.7–4.9bn of profit this year, so it doesn’t need the reactors to work any time soon. It’s also already making money from data centres today, just not from nuclear. Data centres are buying its generators, and orders in that part of the business were up 55%. THE BEAR CASE This is a long game. Wylfa doesn’t get its final go ahead until 2029, and the first power isn’t expected until the mid 2030s. First of a kind nuclear projects nearly always end up late and over budget, and GE Hitachi is already building a rival reactor in Canada. Honestly, if you buy Rolls Royce today you’re mostly buying jet engines. If air travel slows down, the nuclear story won’t save the share price. MY VERDICT I own it, but I didn’t buy it for the reactors alone. I bought it because very few companies on earth can build big jet engines, and airlines pay Rolls Royce for every hour those engines fly. The nuclear side is what could make it a much bigger story over the next ten years, and I like that the jet engines are paying for that bet rather than shareholders. If Wylfa gets the green light in 2029, that’s when I’ll start taking it really seriously.

Anthropic, filed publicly for its IPO yesterday, it's going on the Nasdaq under the ticker $ANT with an arrival of mid november. The revenue growth is impressive: $16.3bn in the first half of this year, up from $1.2bn a year earlier. But the number that shocked me was the valuation. Reports say it’s targeting somewhere between $1.5 and $2 trillion, for a company that lost $80.7bn in the same six months (although about $72bn of that was a one off accounting charge). It’s an incredible business, but it isn’t a bottleneck in the way I usually look at them, as there are other AI labs fighting for the same customers. I never buy IPOs on day one as there's sometimes a period of uncertainty and repricing, it's pretty much just a gamble. $SPCX is a great example. Unfortunately, I do think if this event is not played perfectly we will see a lo tof uncertainty in the markets.

Nearly every US missile and rocket booster needs a chemical called ammonium perchlorate, which is basically the oxygen supply that lets solid rocket fuel burn. And the main American supplier is a plant in Utah owned by $NEU NewMarket, a company that mostly sells additives for engine oil. In its own annual report it says the overseas producers in France, Japan, China and India aren’t qualified for most US military and NASA programmes. It’s also spending $100m to grow capacity by more than 50%. The bit I find funny is that you’d never know any of this from the share price, because it’s priced like a lubricant company.

Didn’t expect to find this one in the FTSE 250. Chemrig owns a factory in a tiny town in Norway, which goes all the way back to Alfred Nobel, and it makes the explosive that goes inside NATO missile warheads. Norway’s Defence Minister literally called it one of only two major makers of this stuff in Europe. It’s been running flat out, and Northrop Grumman has signed a 15 year deal to buy from it. Lower on the bottleneck side of things, but it's a chokepoint nonetheless.

Just sell it all now.

You can only own one for the next 10 years. Which are you taking? $NBIS $IREN $CRWV $MRVL

China makes around 90% of the world’s rare earth magnets, which is why America is pouring billions into companies like $MP to build its own. But here’s the part nobody mentions. IDTechEx expects so much new magnet capacity to come online that by 2037 the world’s factories could be running at just 59% of capacity, which means lower prices and a squeeze on the very companies being built to break China’s grip. Today’s bottleneck can become tomorrow’s glut, and the hard part of investing in one is knowing which stage you’re buying into.

Things I’d tell myself at 17 when I bought my first stock: •If you hate holding it, you should probably keep holding it. •Once you've gathered all the facts, make a decision, any decision. •Don't sell unless something is broken. In short: Shut up, Hold it, Shut up.

People seem to think that more data centres only means more energy use. While that is true, there's a key part their not mentioning. Energy efficiency. Yes the data centres will use lot's of energy but the real money comes from developing better ways to harness the available energy, instead of just making more.

America’s only working commercial uranium enrichment plant is in New Mexico, and it’s part owned by the British government. It’s run by Urenco, which is owned by the UK and Dutch governments and two German utilities, and with Russian imports banned from 2028, it’s one of the few places US reactors can turn to. Russia still holds 44% of the world’s enrichment capacity. It’s one of the most important nuclear bottlenecks in the West, and you can’t buy a single share of it.

Every advanced chip in the world is printed a bit like a photo. Light shines through a stencil, and the pattern gets printed onto the chip. $ASML makes the machines that do the printing, which most people have heard of. But if that stencil has even one tiny flaw, the same mistake gets copied onto every single chip it prints. The only company in the world that can properly check those stencils is a Japanese firm called Lasertec, $TSM $Samsung and $INTC all depend on it. The strange part is that being the only one didn’t stop its shares dropping 14% in a single day in February. Owning a bottleneck protects the business, but it doesn’t always protect the share price.

Everyone’s excited about Rolls Royce building nuclear reactors, but it’s already making money from data centres today, and it has nothing to do with nuclear. Its Power Systems division sells mtu generators, and data centres are buying them for backup and even main power. Power generation orders were up 55% in the first half, revenue up 41%, and the division’s margin jumped from 15.3% to 20.3%. Full Rolls Royce breakdown new week, nuclear included. I own it so I’ll be honest about the bits I don’t like too.

If you want a gas turbine to power a data centre, you might be waiting until 2030. $GEV is effectively sold out through 2028, with only about 10GW of slots left for 2029 and 2030, and Wood Mackenzie expects turbine prices to hit around $600 per kW by the end of 2027, nearly triple 2019 levels. Everyone’s watching the chip shortage, but the power shortage might end up being the bigger story here.

To get to this point at 20 years old I have failed, A LOT. Bought high, Sold low. Predicted the market (and failed). Over leveraged. Over risked. Revenge traded. Blown up accounts. But through all of it, i've become a better trader, investor and oddly, a better person. This has been the best and worst thing I've ever started beacuser it teaches you patience, respect and a lot about yourself. It's the only therapy that pays you to grow as a person.

The first stock I ever bought was Nvidia at around $49, age 17, mostly because I’d heard people talking about it and didn’t really understand what a GPU was. It’s around $220 now. I'd love to say I held it that whole time, but I didn't. The inexperienced investor in me sold and bought a few times. It taught me the most expensive lesson in investing, which is that understanding why a company matters is worth more than any price target. Anyone else’s first ever buy turn out better than they deserved?

Top tier breakdown, definitely worth a read for anyone interested in energy. Keep smashing it @insidegadsmind

The full $FSLR breakdown. THE BOTTLENECK Nearly every solar panel in the world is made from silicon, and China controls roughly 95% of the polysilicon, ingots and wafers that go into them. First Solar doesn’t use silicon at all. It coats glass with a thin semiconductor layer instead, so its whole supply chain sits outside China. That matters a lot, the US has around 60GW of panel assembly capacity but only about 3GW of silicon cell capacity, so most American made panels are still built from imported cells. First Solar makes the entire thing in house, and under the new US rules, it’s one of the very few panel makers developers can be sure qualifies for tax credits. THE BULL CASE → 45.1GW backlog worth $13.6bn, running to 2030 → US factories booked up through 2028, with a good chunk tied to Google and big tech → No long-term debt and $1.7bn net cash → US buyers pay $0.36 a watt, while India pays $0.20 for the same panels That price gap is what a bottleneck looks like on an invoice. THE BEAR CASE → Over 80% of this year’s expected profit comes from the 45X tax credit, which starts phasing down from 2030 → Without tariffs and the anti-China rules, cheaper Chinese panels would win on price → Shares fell 10% in January and 14% in February, followed by a shareholder lawsuit in June → Q2 revenue was down 4% on last year MY VERDICT It’s not really my kind of stock, because I prefer bottlenecks that come from technology rather than tax law. But building a completely separate supply chain while everyone else is tied to China is rare, and I’d rather a business depend on Washington than Beijing. I can see why people own it, but I’d want a lower price first. On my watchlist but I’ll be watching Q3 on 29 October.

This doesn't feel like a big enough pullback to be called a correction, but it's definitely allowing some people to get in on previously missed names. $NBIS $CRWV $IREN To name a few.

Weekly Bottleneck: $FSLR First Solar is the biggest American owned solar panel maker, headquartered in Arizona, and it’s been around since 1999. It doesn’t sell to homeowners, it makes huge panels for utility scale solar farms, the kind that cover hundreds of acres and power entire towns. It’s sold over 100GW of panels worldwide, and it has around 14GW of factory capacity in the US, mainly in Ohio, plus plants in India, Malaysia and Vietnam. What makes it unusual is how it makes them, which I’ll get into tomorrow. Thanks @insidegadsmind for the recommendation

Name a company the world can’t function without that nobody talks about. Not $NVDA $TSM $ASML Something less obvious. Best answer becomes a future weekly bottleneck.