$MU (-4%)

The memory manufacturer Micron has now presented its first SSD with a 5th generation PCIe interface for end consumers. The new connection to the system should ensure significantly higher data transfer speeds.

Four capacity classes

The Micron 4600 comes in the M.2-2280 format and is available in four storage configurations: 512 GB, 1 TB, 2 TB and 4 TB. According to Micron, the new SSD doubles the bandwidth compared to the previous Gen4 SSDs. According to the manufacturer, the sequential read and write speeds reach up to 14.5 gigabytes per second and 12 gigabytes per second respectively, which represents a significant improvement on the previous Micron 3500 model.

The random read and write speeds have also been improved and reach up to 2.1 million IOPS. Data-intensive applications in particular, such as various AI applications and the training of large language models (LLMs), benefit from these performance increases.

In addition to the improvements over previous SSD generations, the Micron 4600 offers up to 11 percent more performance than other current PCIe Gen5 models according to the PCMark10 benchmark, the company explained. And significantly better results were also achieved in the SPECwpc5 benchmarks. However, independent tests of the SSD will have to show whether this is actually achieved in everyday practice.

Soon in finished PCs

In addition to the improved speeds, the new SSD also scores with optimized energy consumption. Thanks to the G9 TLC NAND technology, the Micron 4600 works more efficiently and consumes up to 8.5 watts at maximum performance. In addition, the average operating time is specified as two million hours.

The Micron 4600 PCIe Gen5 SSD is now available for OEM partners worldwide. It should therefore soon be found in high-end computers from various manufacturers that rely on the latest AMD Ryzen 9000 and Intel Core Ultra Series 2 processors. It has not yet been announced when the memory will be available in retail stores and how much it will cost.

Summary

Micron presents first PCIe Gen5 SSD for consumers

The new SSD achieves read speeds of up to 14.5 GB/s

Available in capacities from 512 GB to 4 TB in M.2-2280 format

Improved random read and write performance

Optimized energy consumption thanks to G9 TLC NAND technology

Initially available for OEM partners worldwide

Retail availability and pricing not yet known

https://m.winfuture.de/news/148955

Yesterday I sold my position in Micron with about 10% profit. Compared to others here, that's not a huge amount of course, but you don't have 10% gains every day and so I preferred to play it safe now, especially when I was down over 10% for a month before.

Now I just have to think about where to reinvest the money.

I wanted to share a comment with you that gave me food for thought:

"If the release of a single open source AI model is enough to crash the value of these companies, what does that say about the value they were offering?"

$NVDA (-3.63%)
$AVGO (-3.16%)
$MU (-4%)

Today is shopping with a discount

$NVDA (-3.63%)

$MU (-4%)

these items have been extended today💪

Semiconductor stocks and electricity shares are falling today because of the Deepseek panic. This panic is probably exaggerated. We'll talk about it in tonight's podcast. So be sure to subscribe to the podcast: https://open.spotify.com/show/3SHlvxmpYWkpiWnykajNPa

Semiconductors:

-12% Broadcom $AVGO (-3.16%)

-12% TSMC $2330
$TSM (-0.58%)

-11% Nvidia $NVDA (-3.63%)

-11% BE Semiconductor $SMH (-2.93%)

-11% ARM Holdings $ARM (-3.76%)

-11% ASML $ASML (-0.72%)

-9% Dell $DELL
$DELL (-1.59%)

-9% Micron $MU (-4%)

-6% AMD $AMD (-2.29%)

-5% KLA $KLAC (-2.61%)

Power stocks:

-20% Siemens Energy $ENR (-4.27%)

-15% Bloom Energy $BE (-6.5%)

-14% GE Vernova $GEV (-8.21%)

-13% Vistra $VST (-7.68%)

-11% Schneider Electric $SCHNEIDER

#deepseek
#openai
#ai
#ki
#china
#podcast
#spotify
$SPOT (-3.66%)
#semiconductor
#halbleiter

Today, my overweighting of the semiconductor sector is once again falling flat on its face.

Which stocks would you sell and why? Or would you stick with it everywhere?

$NVDA (-3.63%)
$HY9H (-2.48%) . $QCOM (-4.11%) . $MU (-4%)
$ASML (-0.72%)
$8035 (+0%)

$TSM (-0.58%)
$2330

• Revenue: $26.24B (Est. $25.83B) ; +39% YoY
• Net Income: $11.31B (Est. $11.17B) ; +57% YoY
• Gross Margin: 59.0% (Est. 58.5%) ; +53% YoY
• Oper. Margin: 49.0% (Est. 48.1%)
• FY25 CapEx: $38B - $42B (Est. $35.15B)
• FY24 CapEx: $29.76B (Est. ~$29.5B)

Q1'25 Guidance:

• Revenue: $25B - $25.8B (Est. $24.43B)
• Gross Margin: 57% - 59%
• Operating Margin: 46.5% - 48.5% (Est. 46.4%)
• Management expects a ~5.5% sequential revenue drop (smartphone seasonality) but sustained robust AI demand.

Long-Term Revenue CAGR: ~20% (2024-2028)

• AI-related revenue was mid-teens percent of total in 2024 and is expected to double in 2025, with a ~40% CAGR for AI accelerators through 2029.
• AI & HPC cited as main growth engines
• Smartphone & PC segments also gain from higher silicon content

Q4 Process & Segment Details:

• Wafer Shipments: 3.418M; UP +15.6% YoY
• ASP per wafer: ~$6,850 (FY basis); UP +19% YoY
• Advanced Technologies (7nm & below): 74% of total wafer revenue (vs. 69% in Q3)
• 3nm: 26% (vs. 20% in Q3)
• 5nm: 34% (vs. 32% in Q3)
• 7nm: 14% (vs. 17% in Q3)

Q4 revenue by product platform

• HPC (incl. AI): 53% (vs. 51% in Q3) - HPC up +69% YoY
• Smartphone: 35% (vs. 34% in Q3)
• IoT: 5%
• Automotive: 4%
• Consumer Electronics: 1%
• Others: 2%

Q4 Revenue by Geography

• North America: 75% (vs. 71% in Q3)
• China: 9%
• Asia Pacific (ex-China): 9%
• Japan: 4%
• EMEA: 3%

Capital Expenditure:

• FY24 CapEx: $29.76B (Est. ~$29.5B)

- ~70% allocated to advanced nodes (N3, N2)

- ~10-20% for specialty tech & non-wafer (e.g., advanced packaging, mask)

• Sees FY25 CapEx: $38B - $42B (Est. $35.15B)

- "Higher than 2024" to fund advanced nodes (N3, N2) & packaging expansions

- Overseas fabs contribute to increased spending

Advanced Packaging (CoWoS, SoIC):

• ~8% of revenue in 2024; expected to exceed 10% in 2025
• "Any rumors about CoWoS order cuts are simply not true. We continue to increase capacity."

Comment from the management:

Sales drivers and overall performance in the 4th quarter

• "We attribute our sales performance in the fourth quarter to strong demand for 3 nm and 5 nm process technologies."
• "Advanced technologies (7nm and below) accounted for 74% of total wafer sales compared to 69% in the third quarter."

Full year 2024 and future outlook

• "TSMC expects 2025 to be another strong growth year, with revenue growth of over 20% in US dollar terms."
• "Over the next five years (from 2024), we expect a compound annual growth rate (CAGR) of around 20%, driven primarily by AI-related demand and the continued growth of silicon content in smartphones and PCs."

• "In 2024, AI-related revenue was a mid-double-digit percentage of TSMC's total revenue. It is expected to double by 2025 as the strong increase in AI demand continues."

• "We expect a compound annual growth rate (CAGR) of 40% for AI accelerator revenue over the next five years (2025-2029). This is an important driver for future growth."

AI demand, HPC and HBM

• "Memory will grow overall, but HBM (High-Bandwidth Memory) will grow very fast."

• "Demand for AI has more than tripled at TSMC in the last year and we expect it to more than double again in 2025."

• "We continue to see a robust demand profile in AI and can only hope that we can get enough teams and capacity together to support this growth."

Comment on rumors and expansion abroad

• "Rumors about capacity cuts at CoWoS are just rumors - we are continuing to expand our capacity. There are no plans to cut orders."

• "Factories abroad, including in the US and Japan, have an annual gross margin dilution effect of 2-3%, mainly due to smaller scale and higher supply chain costs. We expect this 2-3% margin effect to continue over the next five years."

• "Taiwan will always be the first to ramp up new nodes due to its proximity to our R&D labs (Hsinchu). Other factories abroad will follow depending on customer demand and government support."

• "Our factory in Arizona has started mass production of N4. The second factory is on schedule to produce N3 or N2 technology. Japan's second factory is planned for 2025 and Europe's first factory is also on schedule."

Non-AI markets and silicon content

• "Smartphones are still experiencing low single-digit growth, but the trend towards higher silicon content (AI features, advanced functions) will drive further growth and shorten replacement cycles."

• "PC is similar: overall unit growth is modest, but AI features in PCs are increasing, leading to advanced nodes and higher ASP."

Export controls and demand from China

• "We believe the new US export rules are manageable. We will apply for special licenses for customers as needed, especially in the automotive, industrial or other non-AI use cases."

• "China remains an important market, but currently only accounts for around 9-10% of sales. We continue to work within the regulatory framework."

Silicon photonics and packaging

• "We've had some technical success with silicon photonics, but we probably won't reach significant volumes for another year to a year and a half."

• "Advanced packaging accounted for over 8% of sales in 2024 and is expected to be over 10% in 2025. Our margins for advanced packaging are still slightly below the company average, but are improving."

• "The demand for CoWoS is extremely high for HPC and AI. Other applications (such as smartphones or PCs) may adopt CoWoS in the future, but for now the focus is on AI."

HBM, partnerships and supply chain

• "We are working with all memory manufacturers to provide advanced logic for HBM controllers. It could be another 1.5 years before significant product production from HBM-related logic makes a big revenue contribution."

• "We do not see TSMC as a bottleneck for AI demand. We are expanding capacity for advanced packaging (CoWoS) as quickly as possible and working with memory partners to align capacity."

Margins & pricing strategy

• "We operate in a capital-intensive business and therefore need a healthy gross margin to continue investing in advanced technologies. Pricing must reflect this."

• "We are seeing several cost headwinds including N3 ramps, N2 R&D, conversions from N5 to N3, inflationary pressures and higher costs overseas. We remain confident that we will achieve a gross margin of ~53% or more in the long term."

• "For wafers produced in the US, we are discussing with customers how they will bear the higher costs. Customers understand that the cost structure is different."

Further questions and answers for CEO/CFO

• "We don't comment on competitors' IDMs, we just say that they are important customers. This business is important to us."

• "AI includes CPUs, GPUs, ASICs and HBM logic for HPC/data centers. Edge AI could create 5-10% more chip area in smartphones/PCs, further driving the adoption of advanced nodes."

• "Currently, SoIC is mainly used in AI applications, but we expect more applications in the future."

• "We expect continued strong growth in the AI sector in 2026, although we have not yet made a specific forecast. Our focus is on ensuring sufficient capacity."

• "We are focused on narrowing the cost gap between factories in the US and Japan, but realistically the scale will remain smaller than Taiwan for years to come."

Final note from the CEO

• "2025 will be another strong year. We expect a 20% increase in sales in US dollars, driven by AI and our advanced technologies. We are ramping 3nm and preparing for N2 and beyond to maintain TSMC's technology leadership."

$ASML (-0.72%) , $ASML (-0.42%) , $NVDA (-3.63%) , $AMD (-2.29%) , $KLAC (-2.61%) , $QCOM (-4.11%) , $INTC (-4.07%) , $MU (-4%) ,

$NVDA (-3.63%)

• NVIDIA ANNOUNCES PROJECT DIGITS PERSONAL COMPUTER PRICED AT $3,000.

• Nvidia $NVDA CEO Jensen Huang just said he thinks self driving vehicles will likely be the first multi trillion dollar robotics industry

• KION $KGX (-0.36%) NVIDIA and Accenture optimize supply chains with AI-powered robots and digital twins (German)

• Uber $UBER (-2.45%) Teams Up with NVIDIA to Accelerate Autonomous Mobility

• $MU (-4%) NVIDIA CEO HUANG: MICRON PROVIDES MEMORY FOR NEW GPUS

• Toyota, Aurora, and Continental Partner with $NVDA NVIDIA for Autonomous Vehicle Revolution

• $NVDA Nvidia Launches Cosmos World Foundation Model Platform to Accelerate the Development of Physical AI

Nvidia Cosmos is a platform of state-of-the-art generative World Foundation

models (WFM), advanced tokenizers, guardrails and an accelerated

accelerated data processing and curation pipeline that enables the

development of physical AI systems such as autonomous vehicles (AVs) and

• robots

Posi is now full