'GPT-5' to be 100x more powerful

Also: Nvidia AI antitrust investigation escalates

Welcome, AI enthusiasts

OpenAI’s next-generation AI model, dubbed ‘GPT Next,’ is expected to be 100 times more powerful than its predecessor GPT-4. Meanwhile, the Department of Justice has reportedly issued subpoenas in a growing antitrust probe over Nvidia's AI dominance, and OpenAI is moving ahead with its plans to build its own AI chips. Let’s dive in!

?

GPT-5 to be 100x more powerful

During the KDDI Summit in Japan, OpenAI Japan CEO Tadao Nagasaki provided insights on the expected developments for GPT-5, also known as ‘GPT Next’. This follows earlier hints from OpenAI executives and researchers about a more advanced AI model.

Microsoft’s CTO Kevin Scott also mentioned similar projections at the Build developer conference, while an OpenAI researcher at the Viva Technology festival referred to "GPT-Next" as significantly superior to GPT-4. It's anticipated that the upcoming model will likely incorporate two new AI systems currently in development: "Strawberry," which focuses on advanced math and coding skills, and "Orion," seen as the potential true successor to GPT-4, leveraging high-quality data generated by Strawberry.

The graph from OpenAI Japan hints at a 100-fold increase in performance, raising questions about whether this refers to the model’s intelligence, computational power, architecture, or learning efficiency.

Interpreting this type of exponential growth can be challenging for humans, but there's no doubt that we are on the brink of some fascinating developments.

Nvidia AI antitrust investigation escalates

The Department of Justice is conducting a thorough antitrust investigation into Nvidia's dominant position as a provider of AI chips.

According to Bloomberg, Nvidia and other companies have received subpoenas, indicating that regulators are scrutinizing whether Nvidia has hindered competitors' entry into the market or unfairly penalized customers who do not exclusively use its chips.

This news led to a significant drop in Nvidia's stock, causing a loss of approximately $279 billion (9.5%) in market value in a single day. However, the stock price still remains more than double its value from earlier this year, largely propelled by the ongoing AI boom.

This sharp decline in Nvidia's stock reflects increasing investor doubt about the growth potential of AI, which has been a driving force behind stock gains throughout the year.

Concerns about slower returns on significant AI investments have also exerted pressure on major players such as Microsoft and Alphabet, leading to a decline in their shares after their July quarterly reports.

This trend might indicate a shift in market sentiment towards a more cautious optimism in the swiftly evolving AI sector. Who knows, we would just wait and see.

OpenAI plans to build its own AI chips with TSMC

OpenAI is ramping up efforts to develop its own AI chips in order to reduce dependency on costly AI servers from Nvidia. The organization has secured production slots with TSMC for the upcoming 1.6 nm A16 process, which is expected to feature backside power delivery technology—a first for a node of its kind.

While this chip is still several years away from mass production, this strategic decision could potentially yield cost and performance advantages in the competitive AI landscape. Reports also suggest potential collaborations with Broadcom and Marvell, both of which are relatively new to these advanced nodes, as well as potential partnerships with Apple, a company with its own AI ambitions.

This development is significant as it aligns with OpenAI's long-standing plans to establish its own infrastructure, including a projected $100 billion supercomputer in collaboration with Microsoft, which may require nuclear power plants to operate.

However, creating its own chips represents a major financial and time investment for OpenAI. Nonetheless, it is clear that OpenAI is actively seeking alternative paths as it seeks to reduce its current heavy reliance on Nvidia.

 Tip of the Day        

The generative AI race is heating up as another Chinese firm, after Kling.ai, has launched a new text-to-video model called MiniMax that delivers phenomenal results. You can try Video-01 for free at hailuoai.com/video by registering with any mobile number, even non-Chinese ones. It’s time to get creative!

Till the Next One.

Peace ? and Love ?