Is the AI bubble real or just dot-com déjà vu?
Henry Jennings from Marcus Today sat down recently with Alex Pollock, founder and CIO of Loftus Peak, to talk AI, tech valuations, the dot-com comparison everyone keeps making, and where the smart money is going next.
Full episode of On the Couch:
Loftus Peak manages $1.5 billion and has been running a disruption-focused strategy for 12 years – well before AI became a household word. Pollock traces the fund's philosophy back to watching Fairfax get dismantled by Seek, carsales, and REA Group. "We don't call it a technology fund, we call it a disruption fund," he says. That framing has taken the fund from narrowband internet all the way through to Nvidia, large language models, and what Pollock believes is a genuinely transformative moment in history.
Is this an AI bubble?
The conversation doesn't shy away from the question. With companies like SK Hynix up 1,000% in a single year, and trillion-dollar companies doubling their market caps in months, the bubble question has to be asked. Pollock's answer is nuanced.
The dot-com crash, he argues, happened because the burn rate of companies was no longer credible – there were no real revenue models, and the internet at the time was narrowband. Pets.com and its peers were spending far faster than demand could be created. This cycle is fundamentally different. "We know that the product is real and that the monetisation of it is real," Pollock says. The question now isn't whether demand exists – it clearly does – but whether the level of capital being deployed is appropriate for the revenue that follows. Nobody knows the answer to that yet. But Pollock is comfortable with the uncertainty.
Where Loftus Peak is putting money to work
Beyond Nvidia – the fund's largest position – Pollock flags three areas he is actively increasing exposure to.
The first is China. Loftus Peak has positions in Tencent, Alibaba, BYD, and CATL, the global battery giant. Pollock makes the case that CATL's opportunity extends well beyond electric vehicles – into home energy storage, data centres, shipping, and aviation. "This has got an incredibly long way to run," he says, adding that the company is not remotely overvalued relative to that opportunity.
The second is biotech, anchored by Eli Lilly and the weight loss drug thesis. Pollock argued for Lilly years ago at a conference when almost no one was listening. His view now extends further – into the link between obesity, downstream disease, and the emerging science around solid tumour remission rates for cancers like melanoma, where five-year studies are showing results that were unheard of a few years ago.
The third is streaming – Spotify, Netflix, and Roku – where the fund has held positions for three to five years and is adding.
The picks and shovels approach
A thread running through the whole conversation is Pollock's insistence on owning the infrastructure of AI rather than betting on specific applications. "We're picking the picks and shovels that make AI up," he says – the companies whose products get sold into transportation, logistics, retail, and every other industry being reshaped by machine learning. It's a way of spreading the AI bet across the whole economy rather than concentrating it in a handful of model providers.
The black swan question
Asked about risks, Pollock is refreshingly blunt. "It's all a black swan, Henry." The Middle East, the rules-based global order, the speed of asset price appreciation – all of it warrants watching. But his broader view is that human beings have a reasonable track record of muddling through, and he's positioned accordingly. Loftus Peak is fully invested, running on the expectation that peak negativity has passed – while acknowledging they could easily be wrong.