Why I Rebalanced from $SKHY into Samsung
I’ve concluded that SK hynix has completely dropped out of the AI leadership group in 2H26, so I rebalanced into Samsung Electronics common and preferred shares. Samsung was already a much larger position, but I took this opportunity to move everything into Samsung.
If I were SK Group Chairman Chey Tae-won, I would likely 1) expand AI businesses across SK Group, and 2) lower HBM prices and push a U.S. IPO of Solidigm to ease U.S. geopolitical issues. However, I’m not sure this would necessarily benefit SK hynix shareholders.
Below is my reasoning for cutting SK hynix. There are no hard facts here, only rumors and inferences, but I think the impact would be significant if they materialize.
1. Low HBM price increase in 2027
There is a rumor that, in negotiations with Nvidia, SK hynix is pricing HBM below its competitors to support SK Group’s AI businesses.
2. Solidigm U.S. IPO
Dilutes SK Hynix share holders pie
3. HBM4 quality issues
In a recent report, Daishin Securities addressed the HBM4 quality issues with SK hynix’s key customer. Analyst Ryu Hyung-geun said, “Some have raised concerns that HBM4 quality issues with a key customer could cause business performance to fall short of expectations, but these concerns are overblown.” He added, “Some problems may have occurred at the back-end stage, but it is hard to view this as an unsolvable risk,” and “In our view, it is not yet the stage to discuss compensation for defects.”
AI capex cut fear is behind us. If the US can't even reach a consensus on this at home, how is it supposed to strike a deal with China?
Last month I wrote about how we can build a positive and safe future for everyone: meta.com/thefutureisforevery…
Every lab has the responsibility and incentive to move at the pace required to train its models safely, and the ability to take its own actions to ensure that happens.
The reality is:
- People won't want to use agents that are misaligned with them and that don't do what they ask, so labs have a strong natural incentive to make their models more aligned.
There is a lot of debate about slowing progress on capabilities until alignment catches up. My view is that trust and alignment are quickly becoming the most important capabilities that will differentiate agents and models. Any lab that doesn't focus on alignment will fall behind.
- Labs face significant liability if their models cause harm, so they have a strong incentive to prevent this as well.
Meta delayed shipping Muse for several months to focus on safety and security. We didn't call for everyone else to do this before we would. We just did it as part of our day-to-day work because it was clearly the right thing for people and for us. I'm proud of the security foundations we've built.
- Engaging independent evaluators and advisors is industry best practice. MSL already does this today in several areas because it helps produce better work. Other labs can just do this too. In general, it would be helpful for there to be a larger and more diverse ecosystem of evaluators.
- Committing the significant majority of compute towards serving people rather than racing towards recursive self-improvement is one of the best ways to ensure we develop this technology safely. Meta has made this commitment and other labs can do this as well.
I believe the key to building a positive future for everyone is maintaining the right balance of power. This is within our power to do.
$AVGO AI revenue guidance came in above expectations, erasing the after-hours decline. FY27: $115B (+98% y/y), FY28: $230B (+100% y/y)
The FY28 AI revenue figure in particular sits 15% above the most bullish sell-side estimate. Like Nvidia's FY27 revenue guidance (+70% y/y vs. 44% consensus), this points to earnings visibility running well ahead of consensus.
Bernstein's $150k Bitcoin Forecast — A Realistic Price Range Derived from Past Institutional Targets
Bernstein forecasts Bitcoin will set a new all-time high of $150,000 by mid-2027 and reach $300,000 at the 2029 cycle peak.
- The rationale: the 40-year era of falling interest rates has ended, and with US sovereign debt at $40 trillion, governments will opt for currency debasement over fiscal austerity — plus expanding ETF and corporate treasury buying.
-Institutional Bitcoin forecasts have always been wrong, so I pulled together the historical cases to derive a realistic price range.
*2020–21 cycle
Simple average target: $289,833 (low $146K, high $600K)
Actual high $68,990 → 23.8% realization rate, −76.2% shortfall
*2024–25 cycle
Simple average target: $189,000 (low $150K, high $500K)
Actual high $126,198 → 66.8% realization rate, −33.2% shortfall
- The narrowing shortfall isn't because institutions got better at forecasting — it's because the target multiple itself shrank from 11x to 2.8x.
- In 2020 they threw out 20x numbers on TAM logic like "if it matches gold's market cap." Since 2024 they've anchored to observable flows — net ETF inflows and corporate treasury buying — which pulled the multiple down.
- Applying this to currently live targets: Bernstein $150,000 (mid-2027), Standard Chartered $100,000 (end-2026), JPMorgan $170,000 → average $140,000. That's 1.8x the current price of $78,400, below even the 2024–26 cycle's average multiple of 2.8x.
- The realistic Bitcoin target range based on historical precedent is $90,000–$130,000.
- If Bitcoin breaks $100k (a symbolic number), the media will start talking it up again and the crowd will pile in with praise, and the price action will reinforce itself reflexively. Don't get drunk on it — sell well into that.
finance.yahoo.com/markets/cr…
This chart is wild. Current AI agent is largely restricted to coding and tech savvy white collar areas.
Humans are the minority user of AI
Agents burn nearly 5x the tokens people do, up 14x since February
Charts of the Week: a16z.news/p/charts-of-the-we…
Are we back?
Key Takeaways from Big Tech Conference Calls — and What They Mean for Semis
Overall, neither great nor terrible. Some comfort in the fact that this wasn't the worst-case scenario semiconductor investors had feared.
Except for Google, management teams were tight-lipped about medium- to long-term capex plans. My guess is they're being cautious because signaling heavy capex can crush the stock price and put them in an unfavorable position in capital markets.
On the positive side, Google and Microsoft cloud revenue growth is accelerating (Google +82%, Microsoft +43% y/y), and they seem confident about AI monetization. Amazon's report tomorrow will probably strike a similar tone.
On the negative side, Meta is a bit ambiguous. They keep weighing the ROI of selling compute externally versus using it internally, and it feels like they haven't picked a direction. That said, given Zuckerberg's strong conviction in his consumer AI agent bet (the next-generation market opportunity beyond coding, currently the main source of AI agent demand), there seems to be little incentive to cut AI investment anytime soon.
Separately from all this, semiconductor share prices look likely to be driven by flows and macro for the time being.
GG
Strategy has sold 3,588 $BTC for $216 million to fund dividends on our Digital Credit securities. As of 7/5/2026, we hodl ₿843,775 in our BTC Reserves and $2.55 billion in our USD Reserves. strategy.com/press/strategy-…
Robinhood retail is coming to Korean stock market
- NH Investment & Securities, one of the biggest local brokerages, is in talks with Robinhood to launch a foreign omnibus account service, having visited Robinhood's U.S. headquarters this week and reportedly reached a meaningful agreement on opening the account.
- The omnibus account structure lets foreign investors trade Korean stocks through their local broker without opening a separate domestic account; a January regulatory revision now allows this without special "innovative financial service" designation, accelerating adoption—Samsung Securities was first to market, launching with IBKR in May.
- Whereas IBKR targets professional and institutional clients, Robinhood's mobile-first, MZ-generation retail base could pull significant Korean-stock demand from the U.S. retail market, as rivals (Shinhan, KB, Hana with Emperor Securities, Kiwoom with Webull) race to lock in large overseas platform partnerships.
$HOOD $EWY $DRAM
news.einfomax.co.kr/news/art…
Too big to fail?
OpenAI has reportedly discussed giving the U.S. government a 5% stake as it seeks to ease political pressure around AI, per FT.
At OpenAI’s reported $852B valuation, that stake would be worth about $42.6B
The proposal would involve a public ownership vehicle, similar to a sovereign wealth fund, and OpenAI has suggested other major U.S. AI labs could hand over similar stakes.
Talks are early, no deal is final, and any structure may require Congress.
