"A stylish woman walks down a Tokyo street filled with warm glowing neon and animated city signage. She wears a black leather jacket, a long red dress, and black boots, and carries a black purse. She wears sunglasses and red lipstick and walks confidently and casually. The street is damp and reflective, creating a mirror effect of the colorful lights. Many pedestrians walk about."

This was the prompt from which Sora generated a video and this is a still from it. Having walked through Tokyo several times at night, I can assure you that this video is unimaginably realistic. The image is so lifelike that I caught myself pondering which neighborhood this video was based on.

Developments in AI are happening at an unprecedented pace. Last week saw some extraordinary new products introduced, with this "text to video" service Sora from OpenAI being particularly impressive. In all honesty, I did not expect that after the spectacular introduction of ChatGPT in late 2022, so much progress would be made so quickly in the development of AI applications. AI expert Michiel Schoonhoven did correctly predict last year that in 2024 the growth of AI would accelerate.

OpenAI amazes again

OpenAI became world famous in a week with the introduction of ChatGPT, still the most popular AI application for the general public. Sora is equally revolutionary; it can handle complex commands and generate videos with different environments, characters, actions and emotions. Whether describing a busy street, a forest or an action scene, Sora can visualize it. The generated videos have impressive resolution and image quality, making them look almost like real footage.

No competition for Spielberg yet

Sora is still under development and not publicly available. It is not expected that complete series or movies will be generated with Sora in the coming year, but for visualizing projects, creation of storyboards and education and training it seems to be immediately applicable.

Competitors such as Runway, Google Lumiere and Stability AI immediately responded with brave statements like "game on," making the consumer the big winner in this new race.

Google is there, but is less noticeable

It seems OpenAI knows exactly when Google is going to introduce a new service, only to launch something of its own in the same week that attracts all the attention. Again this week, Sora got all the publicity while Google's launch of Gemini 1.5 was underexposed. Not that Gemini 1.5 isn't good, but The Verge describes Google's problem as follows:  

"Google and OpenAI are running a breakneck race to build the best AI tool right now, as companies worldwide try to determine their own AI strategy and decide whether to sign their developer contracts with OpenAI, Google or someone else. Only this week OpenAI announced "memory" for ChatGPT, and it seems to be preparing for a product in Web search. So far, Gemini seems impressive, especially for those already within the Google ecosystem, but there is still a lot of work to be done."

After search engines now the AI agent

The excellent The Information had the scoop this week that OpenAI, presumably in partnership with Microsoft's still unpopular search engine Bing, is coming up with a direct competitor to Google's search engine (article unfortunately behind expensive paywall).

Both Google and OpenAI seem committed to developing AI agents for the consumer market. So where you would normally use Google, or OpenAI perhaps soon, to search for a nice restaurant when you're on vacation somewhere, an AI agent would allow you to make instant reservations or order in the country's language, checkout and have the meal delivered. Something like Apple's Siri, but then useful.

Tech industry once again promises to behave, darling

With such developments, AI will soon invade people's lives in intense and intrusive ways. The question remains whether the tech industry will succeed in managing the potential negative consequences of AI better than it did with the rise of social media.

Amid all the jubilation about AI, and I am such a cheerleader, what remains striking is how easily security measures can be bypassed. "Researchers at Brown University have discovered a way to bypass the security measures of OpenAI's powerful GPT-4 system. The trick? Translating harmful clues into less common languages such as Scots Gaelic (or Zulu) before asking the AI for a response.

The findings show that GPT-4 will easily generate dangerous content, such as instructions for explosives or conspiracy theories, when clues are first translated from English. Of the 520 harmful instructions tested, when translated into languages such as Scots Gaelic, it was possible to create problematic content in nearly 80% of the cases, compared to only 1% in English."

Fortunately, Zulus and Gaelic Scots are peaceful peoples. The tech industry promised betterment at a meeting in Munich last week, without clarifying how the promises will be kept. First goal is to ensure that AI will not play a negative role during elections, as no less than half the world's population, four billion people, are eligible to go to the polls this year.

AI delivers prosperity - first and foremost for its entrepreneurs

For a year now, I've been following the biggest seven tech companies, the two biggest crypto currencies and, as a benchmark, the S&P 500, in a section called Spotlight 9. (Please don't look at me about that name, it was made up by ChatGPT.)

The idea behind it was to see the extent to which the tech world and the crypto market perform as an investment compared to the rest of the business world. Last year, crypto was the clear-cut winner, but upon digging a little deeper, something stood out.

First, Bloomberg reported that virtually all of the wealth created in the last year was generated by AI-related companies. So party time at the Huang family, where not only Nvidia CEO Jensen Huang got to add a few billion, but his distant cousin Lisa Su also became worth $1.2 billion as CEO of chipmaker AMD.

Bloomberg sighs, "Two chipmaker billionaires in one family illustrate the magnitude of the craze around artificial intelligence, which has come to dominate the stock market and is responsible for most of the wealth growth of the world's richest people this year."

I also find it remarkable that both were born in Taiwan and studied engineering at top universities in the US; Su has a PhD from MIT and Huang studied at Stanford. They did not get their jobs by being in the right fraternity or sorority, or from one of daddy's golf buddies. The days when Chinese and Taiwanese immigrants started with a takeout restaurant or a laundromat are far behind us. 

The Biden-administration considers AI and the chip industry that powers it so important, that President Biden is considering providing as much as $10 billion in subsidies to Intel. The very chipmaker that has been huffing and puffing in the AI race compared to Nvidia and AMD. Perhaps Intel should see check if there is another smart cousin available in the Huang-Su family.

When comparing the performance of the "traditional" Spotlight 9 with the forerunners in the AI industry, it is only too apparent how much value is placed on AI companies by Wall Street.

If we leave out crypto stocks and AI leader Nvidia from the Spotlight 9, this year Alphabet (Google), Amazon, Apple, Meta (Facebook), Microsoft and Tesla rose a combined average of 6.5%, despite Tesla's 20% decline pulling the average down considerably. That's still not bad, even compared to the 5.5% rise in the S&P 500.

Newly made up: AI Spotlight 9

The overall AI craze in the market only becomes clear when we compare that 6.5% increase to what is happening this year at companies that I have completely subjectively put together in this "AI Spotlight 9."

These companies are either a driver of AI developments like Nvidia and Super Micro, or a big "profiteer" of AI technology, think Palantir and Snowflake, for example. AMD, Broadcom, Crowdstrike, Gigabyte, Microsoft, Nvidia, Palantir, Snowflake and Super Micro were up an average of 48% already this year!

Please note that I do not give investment advice, I just try to follow developments and if I'm feeling bright eyed and bushy tailed on Sunday morning, I try to interpret them as well. These are emphatically not buying recommendations. So much for the public service announcements.

Also new: retail investor can benefit

With the breakthrough of successively the personal computer, the Internet, the smartphone and social media, the big venture capitalists were the winners. They bought shares when the companies were still worth little, as Peter Thiel bought over 10% of Facebook in 2004 for half a million dollars, at a company valuation of $5 million.

Facebook went public in 2013 at a valuation of $104 billion. The exact amounts are unknown but it can be assumed that the investment of half a million dollars, Thiel made over a billion dollars in profit.

Apple exceptional, from $300 to $157,000

The retail investor, the consumer, could only buy shares of Apple, Amazon, Google or Facebook after they became available on the stock market and the biggest gains in valuation increase had already been made. The revolution took place, the value of the company rose, and then the IPO followed.

With Apple being a unique exception: Apple went public back in 1984 but anyone who had bought shares of Apple stock for the price of an iPod (three hundred dollars) on February 18, 2004, exactly twenty years ago, would have made over $157,000 in profit today.

Nvidia is not expected to generate the same returns as Apple, but it is notable that Nvidia and Super Micro are both over 30 years old. All the companies in my completely arbitrary AI Spotlight 9 have been publicly traded for years, but the big increases are happening right now.

Since the first time I wrote about Super Micro and today, the stock has risen from $573 to $803. And when did I first write about Super Micro? Two weeks ago, on Feb. 4. Things are moving fast in the AI sector!

AI-washing, the word of 2024?

Given this extreme stock market performance, it is probably wise that the U.S. financial market watchdog SEC warned against "AI-washing": the frequent use by companies of the word AI in conjunction with their products, when there is little to no actual AI involved. Most companies use as their core the technology of, say, OpenAI, Anthropic or Google, put it in their own wrapper and then issue ranting press releases.

It's like buying a car from the dealership on Friday, repainting it in a different color with your pals over the weekend, and then issuing a press release on Monday that you are a new car manufacturer. The SEC made it clear that it is keeping a keen eye on companies that engage in AI-washing and will even prosecute where necessary.

One company that certainly does not do AI-washing and deserves special attention is the aforementioned Super Micro. The company works with both AMD, Intel and Nvidia and has the knowledge to use those various chipsets to build complete motherboards for all sorts of applications, including AI. That knowledge advantage won't be caught up anytime soon, though dependence on other people's chips obviously remains a difficult issue to manage. By the way, the founder and CEO of Super Micro is Mr. Charles Liang, and it is probably no surprise that he too was born in Taiwan.

Yahoo Finance could not contain itself and headlined: Super Micro is having a super 2024. If you look a little further, the rapid rise appears to have started longer ago, as the last 365 days the stock rose a whopping 773%. This year, SMCI already rose 181% and I suspect this will not prove to be the high point.

And we are less than two months underway in the new year. Or, given the large Chinese contribution to this development, perhaps I should better say we are only a week into the year of the dragon.