AI progress report: insights from the latest earnings season

Bubble or…double?

The major debate through this earnings season was the degree to which AI infrastructure spend is in a ‘bubble’, especially as the ‘Rubicon has been crossed’ with increasing use of debt financing rather than just pure free cash flow. Our view is the fact that the term bubble is being so actively used is in itself a sign of healthy scepticism.

Jensen Huang, CEO of Nvidia, addressed the concerns head-on during the company’s earnings conference call. While he naturally has a significant vested interest in believing the nascency of the AI revolution, his firm continues to blow away every target that he previously set – and which was widely regarded as far too ambitious at the time¹.

This success means the bar is always high for the industry bellwether. Nvidia’s shares were weak following their very strong earnings report, whereas the consensus analyst upgrades for 2027 earnings are c.30%.

Distinguishing between winners and losers is humbling

Throughout this quarter’s earnings reports the market has attempted to distinguish between perceived winners and laggards in the AI arms race. Meta’s AI return on investment (ROI) has come under scrutiny and Oracle is in the penalty box for its more balance sheet-intensive data centre expansion plans.

The challenge here is apparent with the impressive recovery in the Alphabet (parent company of Google) share price. Only a few quarters ago Alphabet was being viewed as an AI loser with its core search business under siege from ChatGPT and the other LLMs. Now Alphabet is being lauded as the ultimate AI winner owning the entire stack of chips, data centres, AI models and apps.

We attempt to maintain a balanced view as the pace of disruptive innovation is rarely linear and the total addressable market is so vast that the huge potential profit pools can accrue to numerous industry participants.  

Compute spend is compounding, powered by innovation

Alphabet’s launch of Gemini 3, its most recent LLM, is another leap forward on the technology curve. This cutting-edge model outperforms those of its biggest rivals, including OpenAI, on most benchmarks. But the key aspect is how it was trained entirely on the company’s own Tensor Processing Units (TPUs), rather than Nvidia GPUs (graphics processing units).

This is notable because TPUs are cheaper and consume less power. This step-change in performance and power management is the latest integration in the relentless path of progress AI pioneers are able to demonstrate. Faster speeds, enhanced penetration and functionality expansion will further drive the ROI potential, which in turn will maintain a low cost of capital for further investment in this technological revolution. 

AI is already everywhere…and its influence is growing

The ubiquity of AI is becoming ever more apparent. References to application adoption were apparent on most company earnings calls this quarter. The end goal of AI is to help all businesses drive increased revenues and augment productivity. This is arguably the most exciting development as the opportunity set of potential investments for us will expand rapidly.

Our investment strategy is ideally positioned, in our view, to take advantage of this with expertise already honed in unearthing the market’s early adopters of AI technology.

‘Old economy’ companies like US retailer Walmart are showing tangible evidence of supply-chain velocity lifts, dynamic pricing and personalised customer experiences.

‘New economy’ businesses like App Lovin, a mobile app marketing firm, are driving growth with low capital intensity by using AI to deliver high ROI advertising inventory to its clients.

We believe that AI implementation has the potential to transform every industry.