Tech Catch Up – GPT-5 & Intel

OpenAI – GPT-5 – What whale?

• GPT-5 has launched but instead of the promised leap to artificial general intelligence, we got an incremental improvement in performance and a tidying up of the user experience making it easier and more intuitive to use.
• OpenAI has finally managed to get GPT-5 to market, but it has suffered under the weight of the high expectations as GPT-5 is somewhat better than GPT-4 in some areas but is best described as more of same.
• GPT-5 scores somewhat better than GPT-4 in some of the core benchmarks, but fell short of Grok 4 on ARC-AGI-2 which OpenAI forgot to mention at the product launch.
• Furthermore, it still demonstrates no understanding of causality as it continues to fail to perform simple tasks that a small child would get right.
• The attached picture is GPT-5’s response to “draw a picture of a person writing with their left hand” and “draw a picture of a person holding a sign saying AGI is imminent. Circle all of the vowels”.
• Despite the shortcomings, GPT-5 offers a much cleaner user experience, which, I suspect, will be as important in competing with the ever-increasing lineup of equally good competitors.
• Instead of a dizzying array of options, users now have a simple prompt and the model then decides how to execute the task to achieve the best result.
• Recently launched functions such as Agent Mode and Deep Research are still available, but I think that this simpler approach will be appealing to the users that OpenAI is trying to attract.
• The net result is that while GPT-5 offers very little to advance the progress to AGI, it does advance OpenAI’s strategy to dominate the emerging AI ecosystem and become a consumer product company.
• This is how it will challenge the likes of Google and Meta Platforms as when it comes to raw AI performance there will not be that much to tell the different players apart.

Intel – Money is not enough

• SoftBank has announced that it will make a $2bn investment in Intel but how this will help get Intel out of the mess in which it currently finds itself remains a complete mystery.
• Intel will purchase new shares directly from Intel at $23 per share, giving it a 2% stake, underlining Mr Son’s commitment to invest in the USA and help advance its move towards advanced semiconductor manufacturing.
• This goes hand in hand with a potential investment by the US federal government, which could see it take a 10% stake in Intel.
• While these investments would certainly help keep Intel alive and buy it some time, they will do nothing to alleviate Intel’s problem of being stuck in a strategic quagmire or stop its competitors from devouring its business.
• Intel’s strategic choice has always been fairly clear, which is to either invest heavily and catch up with TSMC or break the company up and sell off the pieces.
• Under Pat Gelsinger, Intel was investing to catch-up, and while progress was being made, it was costing more than hoped leading to a rapid unravelling of Intel’s financial position.
• This is what caused the board to lose its nerve and the departure of Mr Gelsinger.
• The problem now is that this strategy appears to be off the table and with no sign of a break-up it is unclear what Mr Tan (new CEO) wants to achieve.
• This leaves Intel open to having even more of its market share in PCs eaten by Qualcomm and AMD, and Nvidia and AMD in the data centre.
• This will lead Intel’s customers to lose confidence, which in turn will exacerbate market share loss.
• Consequently, I don’t think that investments from the federal government or SoftBank do anything to alleviate the situation in which Intel finds itself.
• A recovery requires Intel to make a clear strategic choice and execute that choice successfully, and of this I see no sign.
• This is a gift for its competitors, which I think will be able to continue taking market share without having to try very hard.
• There remains no price at which I would want to take a position in Intel.