From Argentina to GPU's to AI in warfare & much more
In this week's edition, we explore how Argentina might go through a drastic change; the current gold-rush with GPU's, HPC energy consumption, AI in warfare,
Argentina’s drastic “stance” in case of a leadership change
When we mention Latin America in general, we refer to almost 9% of the world population and around 6% of the world’s GDP. Amongst the Latin American countries, Argentina is the third largest in terms of GDP, behind Brazil and Mexico. Hence, any significant happenings in Argentina do have an impact on Latin America. Off late, Argentina has been struggling with economic problems for years, including high inflation and a weak currency. In recent weeks, the situation has worsened, with the peso depreciating by nearly 18% and overnight the shops hiked prices by 20% amidst the backdrop of inflation running in triple digits.
On top of this, an “outsider” candidate Javier Milei has won 30% of the vote in the presidential primaries, and his victory has rattled investors because of his declared policies he would adopt if he win’s the elections scheduled to happen later this year:
Replace the peso with dollar as Argentina’s official currency
Scrap capital controls
Close the central bank permanently
These fundamentally drastic potential changes in the policy making has already raised panic levels within the Argentinian economy. The devaluation of the peso and the rise of Javier Milei have rattled investors, and it is possible that the country could face further instability in the coming months. However, it remains to be seen if Milei will be able to win the elections and even if he does, whether he will implement his economic plan that he has been promising as part of his campaign.
And we keep the focus on BRICS this week
Further to our earlier newsletter BRICS in growth path, we continue the coverage on BRICS as the leaders meet in South Africa while we speak about them. This meet has the potential to make a big mark on the world dynamics, if additional countries like Indonesia, Saudi Arabia (and probably more) are ‘taken’ inside the BRICS. If that happens, the impact will be big:
Bloomberg reports that an expanded BRICS group would represent about half of global output by 2040.
The expanded BRICS group would double the share of the G7, a reversal from the turn of this century.
A bigger BRICS would account for almost half of the global population
Will an expanded BRICS challenge the existing world order, dominated by the US and EU? Time will tell, but the numbers can back that up!
The GPU gold rush happening now!
Tech companies are desperate for GPUs this year due to the exploding demand for artificial intelligence (AI) applications. GPUs are essential for AI applications because they can perform complex calculations quickly and efficiently. The AI industry is growing rapidly, and this is putting a strain on the supply of GPUs.
One reason for the shortage is that Nvidia, a longtime provider of GPUs controlling around 80% of the GPU market, has a virtual lock on the market. A fun fact to mention is that Nvidia only termed GPU (graphics processing unit) when it developed the first version way back in 1999. While the GPU’s were designed initially for rendering graphics in real-time, due to the capability of immense parallelism built in GPU’s, as compared to normal CPU’s, these are very vital for AI computing.
After the avalanche created in Generative AI once ChatGPT exploded into the market, the demand for GPU’s have grown exponentially. A major reason for the shortage is that cloud computing companies are also competing for GPUs. Cloud computing companies offer businesses access to GPUs and other computing resources on demand. This has made it easier for businesses to use AI, but it has also put more demand on the GPU market.
There are now longer wait-lists in the market, stretching to even one year in some cases. Some tech companies are taking creative measures to get their hands on GPUs. For example, venture capital firms are buying chips and then offering them to their portfolio companies. Entrepreneurs are also rallying start-ups and research groups together to buy and share GPUs.
LUMI model for HPC energy consumption
High-performance computing (HPC) is the use of computers to solve very large and complex problems. HPC systems are used for a variety of applications, including climate modeling, nuclear power simulations, financial forecasting, etc. Because of the huge computing needs, which has grown exponentially in last few years because of AI, HPC systems are becoming increasingly energy-intensive.
The International Energy Agency estimates that data centers, which house HPC systems, account for about 1.5-2% of global electricity consumption.
This is expected to rise to 4% by 2030.
There are a number of ways to reduce the energy consumption of HPC systems:
One is to use sensors to monitor and minimize the amount of power sent to different parts of the chip, depending on the task being performed.
Another is to make sure that as much of the chip as possible is doing useful work at any given time.
Finally, some tasks can be shifted to specialized chips like the modern days GPU’s that are designed for those specific tasks.
The standard measure of a data center's efficiency is the power usage effectiveness (PUE), which is the ratio of the data center's overall power consumption to the amount of power used to do useful work. A typical data center has a PUE of 1.58, which means that about two-thirds of its electricity goes to running its computers and the remaining third goes to running the data center itself, most of which is consumed by its cooling systems.
However, the latest numbers (source: The Economist) indicate that the PUE value have not improved that much since 2017.
The LUMI supercomputer, the world’s third most powerful supercomputer, is a good example of how HPC systems can be made more energy efficient. It has tens of thousands of individual processors and is capable of performing up to 429 quadrillion calculations every second.
It is powered by hydroelectricity and its waste heat is used to help warm homes in Kajaani, Finland. This means that LUMI actually has negative emissions of carbon dioxide. Maybe, LUMI will show how to be energy efficient to the rest of the world.
AI in military weapons
The United States and China are racing to develop artificial intelligence (AI) for military purposes. Both countries are investing heavily in research and development, and they are looking for ways to integrate AI into their weapons and systems. One of the priorities for both sides is to develop weapons that can find their targets without human help. This could include autonomous drones or missiles that can strike enemy targets without the need for human intervention. Another priority is to develop AI tools that can identify targets from satellite images. This could help militaries to track enemy movements and to plan attacks.
A recent study found that a third of all known AI-related military contracts in the United States and China were for intelligent and autonomous vehicles. This suggests that both countries are focused on developing self-driving vehicles that can be used for military purposes. The United States has also taken steps to restrict the export of AI technology to China. This is an effort to prevent China from gaining an advantage in the development of AI weapons.
Some experts believe that AI could revolutionize warfare, leading to the development of new weapons and systems that are more powerful and efficient than anything that exists today. However, there are also concerns about the ethical implications of using AI in warfare. Some people worry that AI could be used to create autonomous weapons that could kill without human intervention.