The GenAI Arms Race Is Just Getting Started

This story features NEXTDC LIMITED, and other companies. For more info SHARE ANALYSIS: NXT

Selling in public markets and tumbling tech equity valuations are not telling the full GenAI story, as private markets and China power on.

-Public versus private markets; two different AI stories
-Implementation of GenAI is not slowing
-All roads suggest there is money to be made from AI 
-Is China the new AI face of change?

by Danielle Ecuyer

AI’s not going anywhere

While public markets have been captured by the Trump tariffs’ twists and turns, and major AI-related stocks have fallen out of favour, including the not-so-magnificent anymore Mag-7, private markets are showing an insatiable demand for AI, described as no less than a frenzy by the FT, as US venture capital gorges itself on this thematic start-up wave.

US start-ups are raising more cash than at any time since 2021 due to AI, according to data from PitchBook. The firm highlights AI and machine learning’s private equity investment has reached almost EUR3bn in transactions to date in 2025, following on from a record EUR11.4bn in 2024.

In the US, some US$30bn has been invested this year, with an additional US$50bn in the pipeline. Venture capitalists are reported as working on deals with OpenAI, Safe Superintelligence, and Anduril, a defense tech start-up.

The December quarter of 2024 experienced US$80bn in US fundraising after a two-year trough, and the March quarter is expected to be at a similar level.

Major deals in the offing include OpenAI in talks with SoftBank to raise US$40bn to a US$260bn valuation, the largest funding round ever achieved, post-Databricks’ US$10bn raising in 2024.

Anduril is looking to raise at least US$2bn for more than a US$30bn valuation, which is twice the valuation of a funding round in the US summer of 2024.

Monetising AI, what is the upside

Morgan Stanley takes a deep dive into assessing the potential return on investment from capital spend on GPUs and AI and sees “GenAI driving around a US$1 trillion revenue opportunity by 2028 from only around US$45bn in 2024.”

In summary::

-Some US$400bn in GenAI software spend by 2028 from US$16bn in 2024. The US$400bn equals around 22% of total expected software spending in 2028.

-US$680bn in GenAI consumer spending by 2028 versus US$29bn in 2024, with an increase in GenAI e-commerce of US$469bn from under US$18bn in 2024. GenAI advertising is projected to rise to US$197bn from US$28bn in 2024.

-Spending on GenAI autonomous driving is expected to be US$6bn in 2028 from under US$1bn in 2024, and GenAI wearables US$10bn from under US$1bn in 2024.

-The spending and investment include US$280bn in sales in 2028 from US$115bn in 2024 from GPUs and application-specific circuit boards. Other GenAI hardware, networking, and memory spend is expected to grow to US$276bn in 2028 from US$98bn in 2024.

-GenAI will require 98GW of power by 2028 from 12GW in 2024, at a cost of US$84bn over the same period, up from US$10bn.

-Data centre-related labour costs of US$7bn, up from US$1bn over the period.

Companies embrace GenAI 

McKinsey’s The State of AI report shows the use of GenAI has jumped since the beginning of 2024. The consultant’s report showed 71% of surveyed respondents revealed their companies regularly use GenAI in at least one business function, a rise from 65% at the beginning of 2024.

Organisations are most often using GenAI in marketing and sales, product and service development, service operations, and software engineering business functions. The application in these segments is likely to produce the most value, according to previous research.

McKinsey also highlighted companies with more than US$50m in annual revenue are using GenAI throughout more of their operations than smaller peers.

Shelly Palmer from The Palmer Group takes the discussion around AI applications a step further, highlighting a 1,200% increase in traffic on US retail websites from GenAI sources between July 2024 and February 2025, according to Adobe Analytics.

The data showed Cyber Monday traffic rising 1,950% year-on-year, with consumers driving the shift.

Out of 5,000 US shoppers surveyed, 39% had used AI for online shopping, with 53% intending to do so this year. Consumers applied AI for product research, recommendations, deal hunting, gifts, product discovery, and creating shopping lists (who would have thought it).

Shelly highlights engagement is also higher than for traditional sources, with 12% more page visits, 8% more time spend on sites, and a lower bounce rate of -23%.

While conversational AI increases confidence, it has not yet led to higher conversion rates, which remain -9% behind traditional sources, though the gap is closing.

Expectations continue to rise that AI-driven agents won’t just assist shoppers but may start shopping for them. Shelly believes the latest data are just the beginning of this story.

Seeing the wood for the trees

E&P is the latest research house to address concerns around the GenAI trade and investing theme.

The investing thematic has been impacted by negative headlines, the analyst proclaims, such as Microsoft cancelling leases and the release of the DeepSeek R1 model undermining the US narrative. Both are believed to be over-emphasised and giving investors the wrong message.

Touching on each of the major sectors, E&P views demand from the semiconductor sector as remaining strong, noting Nvidia’s latest quarterly result, which showed Q4 data centre growth up 16% sequentially and 93% on a year earlier, including revenue from the latest GPU, Blackwell, ramping up from Hopper.

The analysts also reference the strength from 1Q25 revenues for Broadcom, up 25% year-on-year, and earnings before interest and tax rising 41% annually.

Regarding hyperscalers, E&P expects capex in 2025 to advance to US$322bn (announced from Apple, Microsoft, Amazon, Meta, Oracle, and Alphabet) compared to US$233bn in 2024.

Recent earnings reports showed growth in Microsoft Cloud revenue of 21% on a year earlier to exceed US$40bn for the first time. The report quotes Microsoft:

“We continue to expand our data center capacity in line with both near-term and long-term demand signals. We have more than doubled our overall data center capacity in the last three years, and we have added more capacity last year than any other year in our history. Azure and other cloud services revenue grew +31%. Azure growth included 13pts from AI services, which grew +157% year-on-year and was ahead of expectations as demand continued to be higher than our available capacity.” 

Amazon and Meta made similar upbeat statements on cloud computing and AI. Quote from Meta:

“These are big investments, especially the hundreds of billions of dollars that we will invest in AI infrastructure over the long term. I announced last week that we expect to bring online almost 1 gigawatt of capacity this year. And we’re building a 2-gigawatt and potentially bigger AI data center that is so big that it’ll cover a significant part of Manhattan if we were placed there.”

Global data centre operators also highlighted robust growth. Equinix noted APAC was the fastest-growing geography at 13% in the latest full-year results, and Digital Realty received over US$1bn of bookings, a new record in 2024.

In Australian news, the analysts detail up to a $1.2bn investment from Partners Group, a global private equity firm, to step up GreenSquare data centre into a new-generation data centre platform.

ISPT, owned by IFM Investors, is awaiting approval for a 170MW data centre in North Ryde at a development cost of $1.8bn.

Amazon is reported as increasing its activity in Melbourne, with three different deals close to finalisation or already finalised for over 200MW of capacity for AWS in Melbourne.

CDC has also announced a 210MW data centre development in Western Melbourne to compete with AirTrunk, Digital Realty, and Equinix.

E&P has a Positive rating on NextDC ((NXT)) with a valuation of $27.76. Both Macquarie Technology Group ((MAQ)) and DigiCo Infrastructure REIT also enjoy Positive ratings with valuations of $105.34 and $5.13, respectively.

Megaport ((MP1)) is equally rated Positive with a valuation of $17.59.

E&P’s Positive rating infers the stock is “expected to outperform the S&P/ASX200 over the coming 24 months”.

When geopolitics and AI collide

The arrival of the DeepSeek model has laid bare the growing challenges and divide between the two superpowers, China and the US.

As described by Ninety One’s Global Tech Sector Analyst, Anton de Plooy, DeepSeek’s models are achieving AI performance approaching the US’s top GenAI LLMs, such as OpenAI’s GPT-4, at a much cheaper cost.

The significance and market concerns around DeepSeek, as outlined in the February 6 GenAI update (see below for link), focused on the ability to create AI outcomes at a much cheaper cost using older Hopper GPUs from Nvidia.

Not unlike the advances made by Huawei and the electric vehicle industry in China, the US’s strategy of attempting to restrict advanced US technologies across semiconductors and software has resulted in China embracing its own innovation pathways.

De Plooy highlights this as the law of unintended consequences, with China embedding AI in state-driven industrial policies such as the New Generation AI Development Plan and Made in China 2025.

Ironically, as the new Trump Administration becomes more isolationist with restrictive export controls, the US government “may be accelerating, rather than hindering, China’s AI capabilities.” the analyst argues.

In contrast, Initial concerns around DeepSeek’s cost efficiencies and the flow-on impact on AI infrastructure have not, to date, materialised. 

Spending by hyperscaler cloud providers in both the US and China has re-accelerated, de Plooy explains, as rising production efficiency results in increasing demand, known as Jevons Paradox.

Ninety One is not alone in questioning the US’s technological supremacy in AI development and the narrative of exceptionalism across stocks, US markets, and technology.

For investors, de Plooy argues AI is “no longer a US-centric phenomenon and accelerated adoption creates opportunities far beyond traditional infrastructure plays in the next phase of AI-driven transformation.”

Is AI disruption causing a headache for software incumbents?

China is not the only country or entity disrupting AI innovation.

Nilesh Jasani, Founder of GenInnov, regularly publishes his latest insights on the ever-changing world of GenAI.

In recent updates, Jasani highlighted the disruptive impact of AI on everything-as-a-service models for software delivery to customers due to the copiability of new software features or even entire products, stating these can be replicated by others.

“Now, if a SaaS app unveils a cool AI-powered feature on Monday, a competitor (or an open-source community) can have a comparable feature out by Tuesday. This dynamic erodes one of the SaaS model’s key defenses: uniqueness.”

Jasani suggests GenAI will not be a distinct standalone aspect but rather an integral part of an offering, much like how electricity altered the functionality of machines.

As such, BYD’s introduction of a self-driving system (software) in all its vehicles, including the lowest-priced Seagull model for US$9,900 to US$12,000 (available in China only), makes Tesla’s full-self-driving (FSD) software, priced at US$8,000 or US$99 per month, questionable.

Jasani explains AI capabilities can be embedded into other devices, offering unique selling points. Regarding software, GenAI may drive a shift back toward product-specific services rather than all-encompassing, cross-sector offerings.

Businesses that are reluctant to change and adapt will no doubt face disruption.

For more reading on GenAI including DeepSeek and Robotics:

https://fnarena.com/index.php/2025/02/06/deepseek-fuels-ai-technology-advances/

https://fnarena.com/index.php/2025/01/28/deepseek-more-ai-excitement-for-2025/

https://fnarena.com/index.php/2024/12/12/the-2024-gen-ai-xmas-special-incl-2025-outlook/

Or check out the dedicated FNArena GenAI section on the website:

https://fnarena.com/index.php/tag/gen-ai/

Find out why FNArena subscribers like the service so much: “Your Feedback (Thank You)” – Warning this story contains unashamedly positive feedback on the service provided.

FNArena is proud about its track record and past achievements: Ten Years On

Click to view our Glossary of Financial Terms