Is the AI ​​bubble about to burst?

“Every developer is scrambling to incorporate AI into their products right now, and Canva is arguably the company that needs the biggest boost to stay ahead of the huge AI push from Adobe and Microsoft,” Gilbert said.

“Canva’s AI text generator ‘Magic Write’ seems to have received a less-than-stellar response, so it makes more sense to acquire an established dedicated AI generation platform rather than develop one from scratch. It really all depends on how innovative Canva’s integration of Leonardo’s image generation capabilities ends up being.”

Barb Hyman, a former Boston Consulting Group strategist, leads another of Australia’s most successful AI startups. Sapia.aiThe company, which has raised $17 million from Woolworths and Macquarie, describes its software as an “AI career coach in your pocket”.

“I love the story of a great Australian tech company acquiring another, it’s a beautiful symmetry. I’m happy for them,” she said.

For Hyman, Australian AI companies shouldn’t even try to compete directly with the likes of Google, which employs thousands of people with PhDs in AI.

“There’s another way to be competitive … and that’s to have a unique dataset. We have a proprietary dataset that took us two years to build initially,” she said.

“People who have unique data are going to see exponential value, rather than just trying to compete with Google or Meta using all the same open source data that everyone else has.”

The success of Canva and Sapia.ai, and the extent to which other businesses adopt AI technology, will be critical to the local economy and investors.

Australian Productivity Commission commissioner Dr Stephen King said Australia’s productivity had been flat over the past decade and the adoption of AI was critical to growth in Australia’s healthcare and education sectors.

“AI could be the way for developed countries to escape their current productivity slump,” King said.

“AI is the first general purpose technology that has the potential to radically improve productivity in service-dominated sectors.

“Australia doesn’t invent a lot of new technology. 98% of our business productivity improvements come from adoption, not creating new things. So the big gains in productivity from AI for Australia will come through its adoption.”

Kim Oosthuizen, head of artificial intelligence at software giant SAP, agreed, saying it was important for businesses to keep humans involved when using AI to reduce error rates and “hallucinations,” a term used to describe incorrect or misleading results produced by AI models.

“Most people think of AI as ChatGPT. But in a business context, people don’t realize these tools exist today, we just don’t realize they are AI,” she said. “I think most people don’t know or care what technology is on the back end, or which vendor is responsible for it, they care more about, does it actually work?”

That question of practicality is crucial for U.S. investors weighing the potential of artificial intelligence and preparing for the first interest rate cut since 2020. Shares of Apple, Microsoft, Google-parent Alphabet and Nvidia have all been hammered in the past week, leading some to question whether the AI ​​hype is over.

Nvidia’s stock price has been particularly volatile, with shares on Thursday down more than 25% from their June high of $US140 ($215) a share. The rebound hit the biggest single-day gain The company’s market value hit an all-time high on Wall Street. Its dominance of the market for chips that program generative artificial intelligence is widely seen as a symbol of the huge opportunity in the field — while others see it as overhyped. Nvidia holds about 80% of the market for AI chips.

At the same time, infrastructure costs are soaring across the board. The field of artificial intelligence shares similar characteristics with the field of cryptocurrency in that it relies on massive computing power.

This week, Meta CEO Mark Zuckerberg said while announcing quarterly earnings that the computing resources required to train Llama 4, a large language model, could be 10 times the resources needed to train Meta. Camel 3Future models will exceed that number, he said.

loading

“By the end of this year, Meta AI is on track to be the most widely used AI assistant in the world,” Zuckerberg told investors and analysts on a conference call.

“For now, given the long lead time for launching new inference projects, I’d rather take the risk of building capabilities before they’re needed than when it’s too late.”

Meta, which is spending billions of dollars in the race for AI dominance, said it will buy 350,000 Nvidia H100 graphics cards by the end of 2024 to meet its AI needs.

Meanwhile, Google parent Alphabet last week reported second-quarter capital expenditures of $13.2 billion, up 91.4% from a year ago. Microsoft this week reported capital expenditures of $19 billion for the quarter, up nearly 77.6% from $10.7 billion a year ago.

But not everyone is convinced the price is worth it.

Jim Covello, chief equity analyst at Goldman Sachs, believes that most generative AI technologies are not ready for prime time. He said that tech giants and other companies plan to spend more than $1 trillion on AI capital expenditures in the next few years, but have achieved little so far.

“Despite its high price, the technology is still far from where it needs to be to be practical,” Coviello said in a recent report on artificial intelligence.

“Overbuilding things that the world isn’t using or isn’t ready for usually leads to bad results.”

For Coviello, most technological change throughout history has been about replacing very expensive solutions with very cheap ones. Replacing jobs with extremely expensive technology is basically the opposite of what should be done, he said.

“We estimate that AI infrastructure will cost more than $1 trillion over the next few years alone, including spending on data centers, utilities, and applications,” Coviello said.

“So the key question is: What $1 trillion problem will AI solve? Replacing low-wage jobs with costly technology is the exact opposite of previous technology transformations I’ve witnessed in my 30 years of closely following the tech industry.”

Still, some analysts believe the high valuations are justified. Rhys Davis, CEO and co-founder of InvestorHub, said the average valuation multiple for artificial intelligence companies is about 25 times, while Canva’s valuation multiple is 40 times and mining companies’ valuation multiples are 70 times.

“Saying that AI is ‘overhyped’ ignores the fact that both tech and non-tech companies are being outperformed by valuations,” he said.

“The transformative nature of AI cannot be underestimated. It is a driving force that enables companies to scale and operate at unprecedented levels of efficiency that were previously unthinkable. Of course, critical is the proprietary and defensible nature of the technology ‘behind the scenes.’”

For Coviello, it may take a long time for the AI ​​bubble to burst. In the meantime, AI infrastructure providers may continue to benefit.

“It is certainly still a matter of debate whether AI technology can deliver on the promise that has so many people excited today,” he said.

“The less controversial view is that AI technology is extremely expensive, and to justify those costs the technology must be able to solve complex problems, but it is not designed for that.”