Technology / Semiconductor / Internet Research

Nicolas Baratte

Equity Bottom-Up

<ul>
<li>SK Hynix and Micron stocks were lagging TSMC and Nvidia as it took ~2 years for HBM to suck up enough Commodity DRAM capacity to stabilize the Commodity market</li>
<li>That’s now done, we’re just at the beginning of spectacular HBM growth for at least 2 more years. The reasons are known: density increases, speed increases, dies thinner...</li>
<li>Stocks: keep or buy Micron and SK Hynix. Samsung remains unattractive imo</li>
</ul>


SK Hynix and Micron stocks were lagging TSMC and Nvidia as it took ~2 years for HBM to suck up enough Commodity DRAM capacity to stabilize the Commodity market
That’s now done, we’re just at the beginning of spectacular HBM growth for at least 2 more years. The reasons are known: density increases, speed increases, dies thinner...
Stocks: keep or buy Micron and SK Hynix. Samsung remains unattractive imo


HBM Stocks Will Keep Running (Micron, SK Hynix), It’s Just the Beginning

<ul>
<li>2Q26 (Jul-25) revenue growth 56% YoY, despite a sharp decline in China revenue (-24% YoY). No H20 revenues from China. H20 charges normalize, margins improve, Net Income 5% above Consensus. </li>
<li>3Q26 revenue growth stable at 54% YoY, not including any H20 revenues from China. Upside from H20 at $2-5bn, max revenue growth 68% YoY. 3Q guidance ~3% above Consensus.   </li>
<li>The stock is not expensive. Trading at its average 47x trailing PEx, 33x forward. Very little risk to Consensus expectations over FY26-28. </li>
</ul>


2Q26 (Jul-25) revenue growth 56% YoY, despite a sharp decline in China revenue (-24% YoY). No H20 revenues from China. H20 charges normalize, margins improve, Net Income 5% above Consensus. 
3Q26 revenue growth stable at 54% YoY, not including any H20 revenues from China. Upside from H20 at $2-5bn, max revenue growth 68% YoY. 3Q guidance ~3% above Consensus.   
The stock is not expensive. Trading at its average 47x trailing PEx, 33x forward. Very little risk to Consensus expectations over FY26-28. 


NVDA 2Q26 (Jul-25) Small Beat. 3Q26 (Sept-25) Small Beat. Everything Is on Track, Don’t Blink.

<ul>
<li>ASML has been underperforming for 18 months. Recent downgrades are a delayed recognition of reality. ASML’s management goals for 2030 (10% revenue growth Cagr) are ok (plausible) </li>
<li>But growth is slowing fast in 2H25-1H26. Growth should re-accelerate in 2H26 as the impact of China and  Mature node lower spending, multiple delays (Intel, Samsung) is fading away. </li>
<li>The stock is getting cheaper, but lacks catalysts for another 2-3 quarters. Trading at 25x forward PEx, that’s reasonable given the firm’s intrinsic qualities.</li>
</ul>


ASML has been underperforming for 18 months. Recent downgrades are a delayed recognition of reality. ASML’s management goals for 2030 (10% revenue growth Cagr) are ok (plausible) 
But growth is slowing fast in 2H25-1H26. Growth should re-accelerate in 2H26 as the impact of China and  Mature node lower spending, multiple delays (Intel, Samsung) is fading away. 
The stock is getting cheaper, but lacks catalysts for another 2-3 quarters. Trading at 25x forward PEx, that’s reasonable given the firm’s intrinsic qualities.


ASML Downgrades, Semi Production Equipment Demand

<ul>
<li>April 2025, the US Gvt extended restrictions on Nvidia’s China-specific H20 chip. Restriction also apply to H20’s memory bandwidth. This means SK Hynix HBM and TSMC CoWoS.   </li>
<li>Nvidia new China-specific chip derive from this: a downgraded chip, half a Blackwell, without HBM-CoWoS packaging. Time to design, manufacture, could lead to ~15% downside risk to FY26 Consensus?   </li>
<li>FY26 revenues could land at US$175bn or  -12% below Consensus at 199bn. $175bn implies 35% growth YoY, Consensus expects 52%. The stock trades at “low” 30x FY26 and 24x FY27.</li>
</ul>


April 2025, the US Gvt extended restrictions on Nvidia’s China-specific H20 chip. Restriction also apply to H20’s memory bandwidth. This means SK Hynix HBM and TSMC CoWoS.   
Nvidia new China-specific chip derive from this: a downgraded chip, half a Blackwell, without HBM-CoWoS packaging. Time to design, manufacture, could lead to ~15% downside risk to FY26 Consensus?   
FY26 revenues could land at US$175bn or  -12% below Consensus at 199bn. $175bn implies 35% growth YoY, Consensus expects 52%. The stock trades at “low” 30x FY26 and 24x FY27.


Nvidia’s Jensen Complaining About Export Restrictions to China: How Much Is at Stake? Quite a Lot.

<ul>
<li>Tier-2 commodity Memory is a lousy business, which requires high Capex and R&amp;D, generates very low Free Cash Flows. DRAM is more sustainable than NAND as peak margins are higher. </li>
<li>Kioxia only makes NAND, has a high market share (16%) but is a Tier-2 supplier with a roadmap lagging SK Hynix by ~2 years, and a high exposure to Consumer. </li>
<li>The Memory cycle has recovered, probably close to peak earnings. That’s why stocks have been declining (Micron, Hynix).  Part 2 will cover financials (there's a bad surprise) and valuations. </li>
</ul>


Tier-2 commodity Memory is a lousy business, which requires high Capex and R&amp;D, generates very low Free Cash Flows. DRAM is more sustainable than NAND as peak margins are higher. 
Kioxia only makes NAND, has a high market share (16%) but is a Tier-2 supplier with a roadmap lagging SK Hynix by ~2 years, and a high exposure to Consumer. 
The Memory cycle has recovered, probably close to peak earnings. That’s why stocks have been declining (Micron, Hynix).  Part 2 will cover financials (there's a bad surprise) and valuations. 


What Is Kioxia? And Should You Bother? Part 1: Why NAND Is so Hard (And Not Attractive)

United States

Micron Technology, Inc., through its subsidiaries, manufactures and markets dynamic random access memory chips (DRAMs), very fast static random access memory chips (SRAMs), Flash Memory, other semiconductor components, and memory modules.

Price Gauge
52-week Range	0.00 - 0.00
Market Cap (USDm)	0.00
P/E	0.00
Dividend (Yield)	0.00

Micron Technology (MU US)

HBM Stocks Will Keep Running (Micron, SK Hynix), It’s Just the Beginning

HBM revenue and profit growth set to soar, Micron and SK Hynix stocks recommended for investment. Samsung not as appealing.

Nicolas Baratte

NVDA 2Q26 (Jul-25) Small Beat. 3Q26 (Sept-25) Small Beat. Everything Is on Track, Don’t Blink.

Nvidia strong revenue growth in 2Q26 and 3Q26 despite decline in China revenue, with upside potential from H20 and margins improving. Stock not...

Nicolas Baratte

ASML Downgrades, Semi Production Equipment Demand

ASML's underperformance is a delayed reality check, but management's 2030 goals are plausible. Growth is slowing but will re-accelerate in 2H26....

Nicolas Baratte

Nvidia’s Jensen Complaining About Export Restrictions to China: How Much Is at Stake? Quite a Lot.

Nvidia new China-specific chip GPU will take time, leading to potential downside risk for FY26 revenues. But the stock is trading at low multiples

Nicolas Baratte

What Is Kioxia? And Should You Bother? Part 1: Why NAND Is so Hard (And Not Attractive)

Tier-2 NAND memory supplier Kioxia faces challenges with high Capex and R&D costs, low Free Cash Flows, and lagging behind technology leader SK...

Nicolas Baratte

Price Gauge

52-week Range

Market Cap (USDm)

P/E

Dividend (Yield)

HBM Stocks Will Keep Running (Micron, SK Hynix), It’s Just the Beginning

​HBM revenue and profit growth set to soar, Micron and SK Hynix stocks recommended for investment. Samsung not as appealing.

Nicolas Baratte

NVDA 2Q26 (Jul-25) Small Beat. 3Q26 (Sept-25) Small Beat. Everything Is on Track, Don’t Blink.

Nvidia strong revenue growth in 2Q26 and 3Q26 despite decline in China revenue, with upside potential from H20 and margins improving. Stock not...

Nicolas Baratte

ASML Downgrades, Semi Production Equipment Demand

​ASML's underperformance is a delayed reality check, but management's 2030 goals are plausible. Growth is slowing but will re-accelerate in 2H26....

Nicolas Baratte

Nvidia’s Jensen Complaining About Export Restrictions to China: How Much Is at Stake? Quite a Lot.

​Nvidia new China-specific chip GPU will take time, leading to potential downside risk for FY26 revenues. But the stock is trading at low multiples

Nicolas Baratte

What Is Kioxia? And Should You Bother? Part 1: Why NAND Is so Hard (And Not Attractive)

​Tier-2 NAND memory supplier Kioxia faces challenges with high Capex and R&D costs, low Free Cash Flows, and lagging behind technology leader SK...

Nicolas Baratte

HBM revenue and profit growth set to soar, Micron and SK Hynix stocks recommended for investment. Samsung not as appealing.

ASML's underperformance is a delayed reality check, but management's 2030 goals are plausible. Growth is slowing but will re-accelerate in 2H26....

Nvidia new China-specific chip GPU will take time, leading to potential downside risk for FY26 revenues. But the stock is trading at low multiples

Tier-2 NAND memory supplier Kioxia faces challenges with high Capex and R&D costs, low Free Cash Flows, and lagging behind technology leader SK...