AI reckoning

THE global artificial intelligence (AI) investment boom is entering a potentially dangerous phase, and investors in Malaysian semiconductor, data centre (DC) and infrastructure stocks would do well to reassess their expectations.

The political backlash against AI infrastructure development in the United States, particularly following the midterm elections, coupled with the controversy surrounding Nvidia Corp chief executive officer (CEO) Jensen Huang, has introduced fresh uncertainty into an industry whose valuations have largely been premised on uninterrupted growth.

For the local market the implications could be significant. Malaysian semiconductor companies have enjoyed exceptional demand, expanding order books and strong earnings growth, reinforcing investor confidence in the so-called technology supercycle.

So too have stocks involved in building DC infrastructure, including energy, engineering and construction companies.

Many of these companies have also been posting solid earnings.

The latest selloff in Nvidia has brought this vulnerability into sharper focus.

Its shares tumbled in after-hours trading on Wednesday after reports emerged that the board was preparing to remove co-founder and CEO Huang, raising questions about the company’s leadership and succession plans.

The concern goes beyond the immediate controversy.

Huang has been central to Nvidia’s meteoric rise, having shaped its technology roadmap, strategic direction and position at the heart of the global AI ecosystem over the past three decades.

Unlike companies with clearly established leadership succession structures, Nvidia has no obvious successor to Huang, who has previously said he does not believe in succession planning.

This raises a bigger question for investors: what happens to Nvidia, and potentially the broader AI investment cycle, if Huang is no longer at the helm?

Nvidia’s identity and growth story have become closely intertwined with Huang.

His potential departure could therefore unsettle investor confidence in the company’s ability to sustain its technological leadership and growth trajectory.

Meanwhile, the latest round of US technology earnings suggests that investors are also becoming more selective about which companies stand to benefit from the AI boom.

Rather than indiscriminately rewarding companies with indirect exposure to AI, investors are increasingly distinguishing between businesses that provide supporting technologies and those directly supply the infrastructure needed to power AI development.

Companies such as Palantir Technologies, Cloudflare and Arm Holdings, which provide AI deployment software, Internet security and content delivery infrastructure, and chip-design technology, respectively, represent the broader ecosystem supporting AI adoption.

In contrast, companies such as Super Micro Computer, Dell Technologies and Applied Digital, which supply AI servers, graphics processing unit computing capacity and DC infrastructure, offer more direct exposure to the massive capital expenditure behind AI model training and deployment.

This suggests that the market is increasingly placing a premium on companies with tangible exposure to AI infrastructure spending, rather than simply those positioning themselves as beneficiaries of the technology.

Adding to these concerns is the prospect of tighter regulation of AI infrastructure development in the United States following the recent midterm elections.

The results, which saw Democrats make significant gains in Congress and several key state governorships, have raised concerns about greater regulatory scrutiny of AI infrastructure development.

Among the measures being considered are stricter approval requirements for new DCs, restrictions on construction in certain areas, greater disclosure of energy and water consumption, and requirements for operators to bear a larger share of the costs associated with electricity grid upgrades.

This could complicate the expansion plans of major technology companies and DC developers, potentially delaying projects, increasing development costs and affecting the pace of infrastructure spending.

To be clear, AI’s transformative potential remains intact.

The semiconductor supercycle could continue for several more years, and Malaysia could remain an important beneficiary.

But technological revolutions do not guarantee attractive investment returns, particularly when investors pay excessive prices for anticipated growth.

The greatest risk facing the local bourse’s technology and DC darlings may therefore not be an immediate collapse in orders, but a gradual realisation that today’s extraordinary earnings, ­investment commitments and valuation multiples cannot be extrapolated indefinitely.

The AI revolution may continue. The valuations attached to it, however, may not.