Capex Commitment Coming

Gallium Price ($US/KG)
25 April 2024 : Price $798.40 
1 January 2024: Price $755.80

Source

Texas Instruments

An actioned pack week for those closely following the guidance calls from US reporting season. The gem of the week would have to come from Texas Instruments (TI). A revenue beat for the first time in six quarters has analysts calling a bottom for the chip makers’ fortunes. TI has been investing in manufacturing capacity, increasing capex, and reducing free cash flow for only the second time in 2 decades. North American chip manufacturers have underinvested in chip capacity over the last decade which has seen South East Asian and European capacity expand over that timeframe.

Only in late March 2024, a TI executive is stated that they are transitioning its production of GaN chips from 6-inch to 8-inch fabs (investing in fabs in Dallas and Aizu, Japan) in an effort to offer more competitively priced GaN chips. Jerome Shin (Manager of TI’s Korean subsidiary) pointed out that there has been a shift in the perception of GaN chips compared to silicon carbide (SiC) chips since 2022. While GaN chips were previously considered more expensive, this perception is changing because Texas Instruments is transitioning its production from 6-inch fabs to 8-inch fabs. Producing larger wafers means more chips per wafer, leading to increased productivity and lower costs for GaN chips.

Meta

Acceptance of Generative AI from Wall Street to Main Street is now seeing a substantial commitment of capital from big data consumers and users. As we have widely discussed the pending increase in demand from everything from electricity generation, datacentres, hardware/ software, and associated infrastructure will be significant. Companies like Meta are now calling out the impacts on their earnings calls.

“As we’re scaling CapEx and energy expenses for AI, we’ll continue focusing on operating the rest of our company efficiently. But realistically, even with shifting many of our existing resources to focus on AI, we’ll still grow our investment envelope meaningfully before we make much revenue from some of these new products. I think it’s worth calling that out, that we’ve historically seen a lot of volatility in our stock during this phase of our product playbook, where we’re investing and scaling a new product but aren’t yet monetizing it.”

GaN power chips will also be a vital part of the power ecosystem as power delivery efficiency will become a clarion call for those looking to lock in power supply with little spare capacity by incumbent power providers over the next decade.

US Missile Defence Agency

The Missile Defense Agency of the US Air Defense has finally taken possession of Lockheed Martin’s Long Range Discrimination Radar [LRDR]. This innovative radar system is capable of searching and tracking multiple small objects, including every class of ballistic missiles, over a vast distance, and can operate continuously. 

In an official statement, Chandra Marshall, Vice President at Lockheed Martin for Radar and Sensor Systems, shared insights about the company’s cutting-edge radar system. She confirmed that their system combines the advantages of both high-frequency and low-frequency radars, forging a unique method of detecting, tracking, and distinguishing targets. 

This worthwhile program is a testament to the sustained commitment from the US government towards futuristic S-Band radar, terrestrial radar, and the integration of systems. Examples of such comprehensive technologies are readily visible in Lockheed Martin’s Aegis Combat System, Space Fence, and Aegis Ashore systems. With its state-of-the-art Gallium Nitride [GaN] based solid-state radar, Lockheed Martin utilizes the power of its Open GaN Foundry model that builds on mutually beneficial relationships with core GaN suppliers. 

Let’s break it down – the long-range discriminating radar [LRDR] can track hypersonic targets. Why, you ask? Two words – GaN technology. This technology allows the radar to function at higher power levels, while the S-band radar ensures far-reaching detection and tracking abilities.  

Source: https://bulgarianmilitary.com/2024/04/23/us-acquired-lrd-radar-for-all-classes-of-ballistic-missile-detection/

Nanalyze pose the question; Is Infineon Better than Wolfspeed for EV Chip Exposure?

Power chips are specialized integrated circuits designed to handle high voltages and large currents, making them suitable for power electronics applications such as home appliances, automobiles, and industrial machinery.

Materials like SiC and gallium nitride (GaN) are seen as the future of power semiconductors because they offer better performance, higher efficiency, and higher power density compared to silicon. This is particularly the case in green tech markets like EVs and renewable energy systems.

Source: https://www.nanalyze.com/2024/04/infineon-vs-wolfspeed-ev-chips/

UBS Follow Up

The UBS global mining team have been looking at copper demand from the AI/data center thematic and recently adjusted their views after working in conjunction with the UBS capital goods team.

  • The initial estimate of ~65t Cu per MW of capacity came from a Schneider Electric white paper which they now view as potentially being dated. It’s also difficult to determine if copper intensity has changed materially given AI computing requires 4-5x power of a traditional data center, but the increased scale brings down power intensity per unit of processing capacity.
  • UBS still forecast 15-20% growth in data center capacity (MW of capacity) across 2024e-2025e, but this number could be on the conservative side – the AI part of data center growth is at an early stage and they think data centres can lift from ~1.5% of current copper demand in 2023 to ~3% by 2030e (assuming a 15% demand growth CAGR).

 

The Mining Journal discusses the need for the minerals market to catch up on the implications of the AI boom. Full article in the link below. Gallium gets an honorary mention. 

https://www.mining-journal.com/base-metals/opinion/4200419/copper-minerals-markets-catch-implications-ai-boom