Samsung and Hynix are down - massive buying opportunity!

There is a memory shortage. There will continue to be a memory shortage for at least the next 3 years. The SK Hynix and Samsung price is down 20% this week. 

Planned/announced capacity ramp (HBM/AI-memory related), as reported by recent coverage:

T212 £186k Portfolio update 2026-07-03 + comparison with Fisher Investments

Not much change (-£1k)  overall this week. I used the tech stock dip to buy what I hope are cheaper shares, so I now have little spare cash left now in both accounts.

• Bought £2K of MSFT at $371
• Bought £1k Corning at $254
• Bought £1k META at $598
• Bought £1k Broadcom at $366

Here are recent purchases this week (TOELY was last week).

T212 £186k Portfolio update 2026-06-26 and outlook for the future

This week I saw a drop from £193k at the end of last Friday to £186.5k at the end of this Friday 26th June.

Invest IRR 90% and ISA IRR 42%

I had set some stop loss sells and some of these had triggered. Later on in the week, I bought a few back at a cheaper price.

Here is a list of some of the GIA the trades, with the reason and the current price - this is mainly for my notes, but you may find it interesting - not all decisions were 'good' (green background if price went up after selling)!

Will AI Data Centres fail us?

AI companies seem to be the tech darling still - but will it last if they are not profitable and they cannot be connected up to a 24x7 supply of electrickery?

Ramsdens (RFX) agrees to GBP203 Million takeover by FirstCash

I had posted a few blog articles on buying pawn brokers Ramsdens and  FirstCash because these businesses usually do well when the gold price is high and times are hard (people need cash quickly).

Is Pawn good for you - 19th Sept 2025

Invest in Pawn (gold) - 23rd January 2026

and told you I bought £500 of RFX in my 23rd May 2026 blog

Ramsdens has just shot up 30% due to the news that FirstCash is buying Ramsdens.

Trading 212 £193k portfolio update + tips

A great week for my Trading 212 portfolio, gaining over £10k in a week!

£140k deposited over 2 years shows a £53k uncrystallised profit

The Hormuz  'memorandum of misunderstanding' helped a lot, and also the US stock market being closed on Friday helped too!

I have set up stop sells on many of my ISA stocks so that should the market fall on Monday, I will try crystalise my profit. I will buy back in when prices are low again.

Invest GIA (IRR 104%)

ISA (IRR 53%)

I made a few minor trades during the week. The main one was to sell Sandisk at $1977 for a quick £60 profit as it looked like it was falling (but in fact it continued to rise another 1%).

Is Tech now overpriced?

We have more AI IPOs to come and the picks and shovels companies (Micron, Hynix, Seagate, WD, Intel, AMD, nVidia, etc.) seem to be back in favour as many people are beginning to think that AI companies and companies who have invested heavily in AI will have problems actually building the data centres, powering them and especially making a profit from them.

There must be an awful lot of memory, disks and CPUs sitting in warehouses that are destined for data centres but they may never actually get used. Higher-end CPUs/GPUs will be losing value as they are superseded by more recent hardware which is cheaper to run and requires less infrastructure.

Large companies who are currently buying AI as SaaS (and paying hefty subscription fees) are now beginning to want to build their own AI servers so that they are in control of their own AI service and cost constraints.

Companies like HP, Dell and Lenovo may be able to pick up the unutilised AI data centre hardware components quite cheaply and sell them on into the retail enterprise server market.

I am very worried that pure AI SaaS providers will never be profitable and they are asking the retail investor market to fund their folly and then sell their shares and leave the retail investors holding the baby. But I could be wrong.

I don't think the large volume AI users care if they use Deepseek, Claude, ChatGPT or Gemini - what they do care about a lot, is the 'cost of compute'. I think the capital costs of AI will come down and the energy costs will come down too as better, cheaper and more efficient hardware is developed. At the moment, the AI 'dealers' are giving us free or cheap, highly addictive drugs. Once we are hooked, we will then become dependent on them. There will be drug wars, but in the end, it will be a price competitive 'commodity' market.

Packaging is the bottleneck!

Although silicon chip factories can produce more wafers, the chips still need to be packaged. Apparently, there is a big bottleneck here, as most chip packaging is done offshore.

Let us look at a few packaging companies...

Amkor has grown 358% in a year and shows no signs of slowing down!

ASE (ASX) can be bought only in a T212 Invest account - but it has shown similar growth.

1. Foundries & IDMs (Integrated Device Manufacturers)

• Taiwan Semiconductor Manufacturing Co. (TSMC): Dominates the market; Nvidia alone has reserved roughly 60% of TSMC's CoWoS packaging allocation.
  • Capacity Increase: Growing advanced packaging at a massive 80% compound annual growth rate (CAGR). TSMC is aggressively ramping up two new specialized packaging sites in Taiwan and is establishing an advanced packaging site alongside its Arizona foundry complex to support localized US assembly by 2027. [1, 2, 3, 4, 5, 6]
• Intel Corporation (INTC): Operates extensive internal packaging setups in New Mexico (US), Kulim, and Penang (Malaysia).
  • Capacity Increase: Intel is expanding packaging availability to external customers and is qualifying its EMIB-T packaging tech for massive client volume ramps through the latter half of 2026. [1, 2, 3]
• Samsung Electronics (Samsung): Launching a complete "turnkey" ecosystem that bundles memory, foundry nodes, and proprietary I-Cube advanced packaging together.
  • Capacity Increase: Building out entirely new HBM packaging infrastructure across South Korea, Taiwan, and Singapore, which are scheduled to scale aggressively through 2026 and 2027. [1, 2]

2. Pure-Play OSATs (Assembly & Test Specialists)

• ASE Technology Holding (ASE): The global OSAT market leader, controlling a 15.2% revenue share of advanced packaging via its robust 2.5D and 3D solutions.
  • Capacity Increase: Expanding localized footprints and working in close technical lockstep with TSMC to absorb downstream "on-substrate" chip assembly. [1, 2]
• Amkor Technology (AMKR): A premier US-headquartered OSAT provider focusing on high-density fan-out (HDFO) packaging.
  • Capacity Increase: Amkor has tripled its global capital expenditures budget. Its massive new manufacturing campus in Peoria, Arizona broke ground late last year and is accelerating tool installations to begin active production by mid-2027 to package US-fabricated chips. [1, 2, 3, 4]

3. Equipment Suppliers (The Hidden Enablers)

• ASML Holding (ASML): Historically a lithography leader, ASML is seeing an aggressive revenue surge from selling tools tailored toward advanced packaging and heterogeneous integration.
  • Capacity Increase: Ramping up production across two new US manufacturing locations and breaking ground on a major campus expansion in Eindhoven to push out backend tools faster through 2027. [1, 2, 3]
• ASMPT (ASMPT): A major listed supplier in Singapore specializing in advanced die-bonding systems.
  • Capacity Increase: Booking out its manufacturing pipelines through 2027 as foundries struggle to acquire the high-precision hybrid bonders necessary for modern 3D stacking. [1, 2, 3, 4]

2. ASE Technology Holding Co. (ASX) — The Pullback Leader [1]

• The Rationale: ASE is the definitive giant of the OSAT landscape, executing roughly 45% of total elite backend revenue. Because it is a massive structural asset, it has absorbed incredible capital inflows. However, it recently experienced healthy, brief technical pullbacks, taking the short-term froth out of its price. [1, 2, 3]
• Moving Averages: It dropped below its short-term 20-day exponential structures, turning neutral on basic technical indicators like the Relative Strength Index (RSI). This localized "cooling off" makes it an incredibly attractive buy point for long-term investors entering the cycle. [1, 2]
• The Play: Accumulate shares immediately as it validates technical support. It offers the safest institutional-grade exposure to TSMC's downstream overflow. [1, 2]

2. Amkor Technology, Inc. (AMKR) — The Sovereign Security Premium [1]

• The Rationale: At a trailing P/E of ~51.3x, Amkor is certainly not cheap historically. However, it is fundamentally underweighted when considering its unique geopolitical position as the premier US-onshored advanced packaging facility. [1, 2, 3]
• Moving Averages: Amkor features strong momentum, trading roughly 26% higher than its 50-day SMA and nearly 89% above its 200-day structural baseline. While this signifies a clear extension, the aggressive volume profile suggests institutional accumulation. [1, 2]
• The Play: This is a "growth at a reasonable price" narrative within the AI architecture sphere. Rather than buying a full position all at once, utilize a dollar-cost averaging approach to build exposure over the quarter to protect against near-term macro sector volatility.

---

I will track these two (ASX and AMKR) and dollar-cost average into them over the next month or so. I will use top-sliced profits from other shares to provide the cash.

AMAT - Financial Impact of the ASMPT NEXX Buyout

Applied Materials (AMAT) recently purchased $120m of older packaging technology from ASMPT. 

Wall Street has largely looked past any minor near-term drag. Analysts see the integration of NEXX's large-format electrochemical deposition (ECD) tools as a structural growth catalyst. It expands AMAT’s total addressable market within the $3.5 billion advanced packaging equipment sector, positioning the company to capture larger, high-margin package deals for 2027 server designs

My investment in AMAT seems to be paying off (88% gain), hopefully there is still more growth to come...

My thematic investing in 'picks-and-shovels' tech companies has so far paid off, let's hope there is no crash to follow!

Investing in SpaceX via Scottish Mortgage Investment Trust (SMT)

Instead of buying SpaceX, I invested in the Scottish Mortgage Investment Trust fund SMT.

My SMT holding bought mid-April 2026.

SMT had an approx. 20% stake in SpaceX as a private company before the SpaceX IPO.

SMT's Portfolio

Swing trading is easy (in hindsight) but here is how ChatGPT can help!

I opened an ISA on Trading 212 and put £20k in it in April 2025 and started buying stocks. I bought a mixture of individual companies and ETFs. I tried to chase trending stocks. The idea was to use my ISA for more risky stocks which had a higher potential gain (or loss!). In April 2026 I added another £20k and bought more shares.

To date, my T212 ISA has had £40k deposits in 14 months, is at £48.6k right now and an IRR of 42-47% (depending on Hormuz daily variations). 

The question is, could I simply have bought an index fund and done any better for a lot less stress, time and effort?

Trading 212 184k portfolio update

After a disastrous Friday last week, my portfolio has climbed back a bit.

I had some uninvested cash in my T212 accounts, so I took the market dip as a buying opportunity and nearly all £140k of my deposits are now invested.

My all time high was £191k, so my portfolio has not quite fully recovered.

My purchases at the beginning of the week to pick up 'cheap' stocks included:

LLY

META

CME

INTL

MRVL

HPE

AMAT

AVGO

OSCR

SNDK

CSKR

I sold Zevra Therapeutics for a £500 profit (approx 30%) and Elevancc Health for £100 profit.

Portfolio holdings

ISA (£48k - £40k deposited)

OKLO is my biggest loser here. If I sold OKLO now, I would take a £1.1k loss inside my ISA (43%), but unless there is any sudden government investment news, I don't see it improving much. 

Investing in nuclear is a long-term bet and OKLO seems to be just one of many horses in the race but it is backed by Sam Altman (approx $800 million). OKLO is only projected to become profitable in 3 years time. It's share price may go up by 20% this year if I am lucky but most likely it will continue to drift down. OKLO has plenty of cash but is dependent on government energy schemes and a single contract with META. I am going to hold on to OKLO at least until early July for any news of progress (source). Some good news could give a big boost to the share price.

GIA (£136k - £100k deposited)

My losers are mainly gold and silver miners, esp. Lundin Gold and Equinox Gold and OKLO.

I am hoping that the gold miners reports in August will see an increase in their price.

Best performers

Here are some of my best performers in my GIA

I have already top-sliced some of these and my other good performers (MU, SNDK, WDC, AMD, Hynix, etc.).

It may look like I am £36k up in my GIA, but I have crystalised about £10k of gains and HMRC will want their share of these gains next year (minus any tax loss harvesting) - so it is -£2-3k worse than this after deducting taxes!

Note: This is not investment advice and the stocks above have been bought over a period of 1-2 years and may not be such good investments now.

Trading 212 179k portfolio update 2026-06-05

We saw a rapid drop on Friday 5th June 2026 for Tech. stocks including CPU and memory companies. At the end of last week (end May), my portfolio was at £189k so it has lost £10k.

I had set some limit buys and stop losses and most of them triggered. I don't normally do that because it lets the broker (or their agent) see my limits and deliberately stop me out, but this week, because of the insane price rises on some stocks, I wanted to be prepared for a sell-off.

Here are my buys and sells for the first week in June (profit shown for sells in ISA and GIA)...