Clues to the future of AI? Deutsche Bank research lists THESE five themes to watch | Mint

Deutsche Bank, in its research report on artificial intelligence titled ‘Latest in Al: Five clues to the future of Al in recent headlines’, has listed the themes to watch for in the space.

Noting that AI is “as hot as ever”, with AI and ChatGPT still dominating Google search trends, the report also recalled older trends that saw a similar rise in popularity before fading — virtual reality (VR), metaverse, and cryptocurrency.

However, it further said that searches for Al in September 2024 “are as high as they have ever been, and 20 times as high as crypto” and added that five themes “give strong clues” about Al’s future for the rest of the year.

Five themes to watch

1. Al is the elephant in the stock room
2. Data centres are Al Ground Zero
3. Al is still a magnet for funding
4. Political appetite is growing for Al regulation
5. Real-world uses have yet to emerge at scale

First: Al is the elephant in the stock room

The report noted that AI talk in the news is often linked to company stock performances. It said that while ‘Magnificent Seven’ tech stocks have zoomed and now account for 30 per cent of the S&P 500’s total market cap of $50 trillion, it is undecided if the big AI bets are paying off.

The S&P’s rollercoaster ride this year, with 14 per cent growth in H1FY25 followed by a flat Q3, has been driven by the big tech companies, including Nvidia, which has heavily banked on its AI chips. It noted that while investors count on valuations, they do look to long-term metrics to support the share price.

“Yes, shares in the Magnificent Seven have outperformed the S&P 500 for 15 of the past 21 months …but their price-to-earnings ratio is in the mid-30s, far below the 90 times earnings ratio reached by Microsoft and 160 times by Oracle at the time of the dot-com bubble,” the report said.

It also noted that returns are not uniform. The return on invested capital ranges between mid-teens for the laggards, around 30 per cent for the majority, and over 100 per cent for Nvidia (the top winner), raising the question of “overinvestment.”

Alphabet CEO Sundar Pichai told analysts in July the risk of underinvesting is “dramatically greater” than overinvesting, and Meta’s Mark Zuckerberg, in a Bloomberg podcast, acknowledged that hindsight may reveal overspending.

Concerns also exist that Nvidia is fully dependent on “a handful of companies for sales” and that the technology’s capability, value and reliability are not yet established.

Second: Data centers are Al Ground Zero

Fully functional AI needs supportive physical infrastructure such as data centres — more than 10,000 exist worldwide, and each has thousands of servers running the nebulous “cloud”. The report noted that Blackstone and the Canadian Pension Plan Investment Board’s $2 trillion capex plan in the space demonstrates the demand.

This, in turn, will also skyrocket the demand for energy, as bigger and better AI programmes will require larger-scale models and processors to improve and support the performance of various AI tools.

Examples? Dario Amodei, CEO of Amazon-backed Anthropic, the maker of the Claude 3 chatbot, said that training today’s most advanced AI models will cost around $1 billion and that training the next generation will cost 10 times more. In terms of GPUs, Meta’s latest large language model (LLM) Llama 3 was trained on two clusters of 24,000 Nvidia H100 chips and Zuckerberg recently said Llama 4 will need ten times as many.

While energy demand from data centres has been stable at around 200 TWh over the past decade or so, the requirement is now surging. According to the International Energy Agency, Amazon alone bought more than 20 GW of renewable energy from 2010 to 2022, while Meta and Microsoft each bought over 10 GW. Earlier this year, Microsoft said that its carbon emissions have risen nearly 30 per cent since 2020, mainly due to constructing data centres, though it aims to be carbon neutral by 2030.

Three: Al is still a magnet for funding

Despite the doubts, AI is still attractive to investors and draws “generous funding” in the private markets. The biggest winning example for this is OpenAI, which was founded with 10 employees, private equity and a goal to develop safe Al systems that surpass human capabilities.

The money is still flowing. In August, OpenAI said it is looking to raise billions of dollars at a valuation of more than $100 billion; another example is Tokyo-based Sakana Al, founded only in July 2024, it achieved unicorn status ($1 billion valuation) last week.

While global private investment in AI slipped to $96 billion in 2023 overall, infusion in strategic sectors has increased — $18 billion was invested in AI infrastructure, research and governance last year, up from $1 billion in 2022.

Four: Political appetite is growing for Al regulation

Calls for AI regulation are getting louder. California has a bill ready — called ‘SB 1047’, and the European Union has its AI Act. Notably, California, which houses Silicon Valley, hosts 32 of the world’s 50 leading generative Al companies.

The report said SB 1047, passed by state legislators on August 29, reflects “a growing political appetite worldwide to recognise the benefits while, in the words of the bill, avoiding the most severe risks and guaranteeing access.”

SB 1047, which contains provisions obliging companies operating in California to implement a series of safeguards before training advanced Al models, has been met with criticism and approval. OpenAl, Google and Meta have voiced opposition, while Anthropic, Musk and Al-godfather-turned-conscience Geoffrey Hinton gave it their blessing, the report noted.

The EU’s AI Act, imposed on August 1 and the majority of provisions effective in two years, is the world’s first horizontal legal framework regulating AI.

On the global stage, the report noted that, in the absence of specific Al regulation, a cascade of antitrust and copyright cases is shaping the environment. There are also fears about AI replacing jobs, affecting industries, around safety, and the tech being misused.

Five: Real-world benefits have yet to emerge at scale

Overall, despite early adopters, AI has yet to demonstrate its real-world benefits on a large scale, as per the report.

“The gap between high experimentation and low uptake is particularly striking in regulated industries such as financial services and healthcare. In theory, they have the most to gain from accessing and analysing mountains of unstructured data, but in practice, they also have higher stakes than most,” the report noted.

It also acknowledged that generative AI is “certainly flawed” and that questions remain over whether productivity ends justify the investment means. It also noted that historically, technology takes years to apply and often depends on adjacent innovations.

The report is, however, optimistic. “Al does not need to be perfect today to be revolutionary,” it stated.