Big Tech bets bigger on AI: What is fuelling investments amid DeepSeek’s thrifty success?

The vast majority of Amazon’s annual spending of $100 billion will go towards building capacity to support AI workloads on AWS. Only Apple stands apart with its slow-and-steady approach to AI.

These financial results come at a time when AI development is at a crossroads. The AI spending plans of tech companies have been met with fresh skepticism owing to the early success of Chinese AI startup DeepSeek, which claims that its R1 model (rivalling OpenAI’s o1) took fewer resources and less time to build than what was previously believed.

Amazon

Releasing its fourth quarter earnings report, Amazon said on Thursday, February 6, that it plans to increase its annual capital expenditures (capex) from roughly $83 billion in 2024 to $100 billion in 2025.

“The vast majority of that capex spend is on AI for AWS,” Amazon CEO Andy Jassy said on a call with investors. “We spent $26.3 billion in capex in Q4, and I think that is reasonably representative of what you can expect in an annualized capex rate in 2025,” he said.

Shares fell more than four per cent after the company reported its Q4 2024 earnings. (Image: Reuters)

Jassy attempted to justify the jump in AI spending by calling it a “once-in-a-lifetime type of business opportunity.” “We also have capex that we’re spending this year in our stores business, really with an aim towards trying to continue to improve the delivery speed and our cost to serve,” he said.

Aside from its spending plans, the company reported mixed results for the fourth quarter with sales projected to be weaker than expected for the current period.

Amazon has been investing heavily in data centres, networking equipment, and other hardware to meet the growth in demand for generative AI. Last year, the big tech company also announced its own set of foundational AI models called Nova, along with AI shopping assistant Rufus, custom chips, and an AI tool named Bedrock for developing applications on its cloud computing platform at its annual AWS re:Invent conference.

Alphabet

Reaffirming its commitment to AI expansion, Google parent company Alphabet Inc said that it plans to boost its capital expenditures for 2025 to $75 billion, with a $16-$18 billion investment allocated for the first quarter. Its  spending figure for the previous fiscal year was $52.5 billion.

While Alphabet reported that its net income for Q4 2024 increased by 28 per cent from $20.69 billion to $26.54 billion, the company’s cloud revenue increased by 30 per cent from last year to $11.96 billion, but missed analyst estimates of $12.19 billion.

Shares fell more than nine per cent after the company reported its Q4 2024 earnings. (File Photo)

Alphabet’s largest investments in the past quarter were towards AI infrastructure such as servers and data centres to support the growth in business across Google Services, Google Cloud, and Google DeepMind, Anat Ashkenazi, the company’s chief financial officer, said on the earnings call.

She also said that the company will be “bringing more capacity” online throughout the year in order to meet the growing demand for compute power.

Google hopes to capture more of the AI market by reducing costs. The company said that over 4.4 million developers are using its Gemini AI models. “Our obsession with cost per query, I think, sets us up well both to serve billions of users across our products and on the cloud side,” CEO Sundar Pichai told investors on the conference call.

“Part of the reason we are so excited about the AI opportunity is we know we can drive extraordinary use cases because the cost of actually using (AI) is going to keep coming down, which will make more use cases feasible. And that’s the opportunity space. It’s as big as it comes, and that’s why you’re seeing us invest to meet that moment,” he said.

Meta

To fuel its AI strategy, Meta announced that it plans to invest $60-$65 billion in capital expenditures this year, with the company’s total expenses for 2025 estimated at $114-$119 billion. It reported capex of $38.4 billion for the full year of 2024.

A bulk of the spending is related to infrastructure costs, Meta said. It is also looking to hire more employees for work related to AI infrastructure, monetisation, regulatory compliance, etc.

Meta reported its quarterly earnings on January 29, when technology markets were still reeling from the massive sell-off sparked by the news that DeepSeek built a cutting edge, open-source large language model (LLM) at a fraction of the cost of its US rivals.

However, Meta CEO Mark Zuckerberg tried to downplay the impact of DeepSeek’s purported breakthrough on the company and the broader tech industry. “It’s probably too early to really have a strong opinion on what this means for the trajectory around infrastructure and CapEx. […] There are a bunch of trends that are happening here all at once,” Zuckerberg said on a call with analysts.

Shares rose nearly two per cent after the company reported its Q4 2024 earnings. (Image: Reuters)

“It’s possible that we’ll learn otherwise at some point, but I just think it’s way too early to call that, and at this point, I would bet that the ability to build out that kind of infrastructure is going to be a major advantage,” he added.

Monthly active users of the Meta AI chatbot crossed 700 million this quarter, up from 600 million in December, according to chief financial officer Susan Li. The company plans on reaching one billion Meta AI users within 2025.

Meta has been positioning its Llama series of AI models as an open-source alternative to LLMs rolled out by OpenAI and Google. It has claimed that DeepSeek’s emergence validates Meta’s open-source approach to AI development. However, experts and tech enthusiasts have pointed out that Meta’s AI models do not meet the widely accepted definition of open-source.

Microsoft

In fiscal 2025, Microsoft plans to invest $80 billion for the construction of data centres that can handle AI workloads. It is a jump in AI spending from the nearly $50 billion in total capital expenditures in 2024.

More than 50 per cent of these data centres and other AI infrastructure will be located in the US. The software giant has also announced a $3 billion commitment to expand its AI and cloud capabilities in India over the next two years.

Microsoft reported a 31 per cent growth in revenue from Azure and other cloud services. The total revenue from its intelligence cloud division which houses Azure was logged at $25.5 billion, slightly lower than the $25.83 billion estimate.

Amy Hood, Microsoft’s chief financial officer, said that Azure AI services grew by 157 per cent but overall sales lagged as the company does not have enough data centre capacity to meet growing customer demand.

On the impact of DeepSeek’s low-cost AI model, CEO Satya Nadella said, “As AI becomes more efficient and accessible, we will see exponentially more demand.”

Shares declined by seven per cent after the company reported its Q2 2025 earnings. (Image: The New York Times)

According to Nadella, when the cost of cloud computing drops, the prices of inference computing fall too. This will allow customers to consume more and develop more AI apps, he said.

“When I reference these models that are pretty powerful, it’s unimaginable to think that here we are in the beginning of ’25, where on the PC you can run a model that requires pretty massive cloud infrastructure […] So that type of optimization means AI will be much more ubiquitous, and so therefore for a hyperscaler like us, a PC platform provider like us, this is all good news as far as I’m concerned,” Nadella explained.

Microsoft has invested more than $13 billion in OpenAI and provides cloud infrastructure to the AI startup that is behind ChatGPT. It has also integrated OpenAI’s models into Windows 365, Teams, and other products. The company spent the last year rolling out several Copilot AI solutions aimed at enterprises and developers.

Apple

Apple forecast sales to be above expectations for 2025, indicating that the company is confident of the recovery of iPhone sales as it gradually rolls out Apple Intelligence features. It reported capital expenditures of $2.94 billion in the first quarter of fiscal year 2025.

Despite an 11.1 per cent decline in sales in China, Apple reported stronger-than-expected sales of iPads and Macs overall, as new chips reportedly helped persuade customers to upgrade.

The Cupertino-based tech giant has not been keen on pumping billions into establishing AI infrastructure and has, instead, opted for a conservative outlook on AI. Its approach focuses on partnering with other tech companies and integrating their AI technology into its own hardware.

Apple has worked with OpenAI and Google for the current set of Apple Intelligence features.

Shares rose 3.14 per cent after the company reported its Q1 2025 earnings. (Express Image)

“During the December quarter, we saw that in markets where we had rolled out Apple intelligence, that the year-over-year performance on the iPhone 16 family was stronger than those markets where we had not rolled out Apple intelligence,” Apple CEO Tim Cook said on the earnings call.

Unlike Microsoft, Apple has never focused on developing enterprise solutions and has always been in the consumer space. As a result, its short-term profits are not directly tied to AI progress.

The emergence of DeepSeek and its more efficient, less capital-intensive AI models could be good news for the iPhone-maker as it might enable the company to build new features quicker.