AI Spending To Exceed A Quarter Trillion Next Year
Big Tech’s investment in AI is still growing at a dizzying rate, AI Spending To Exceed A Quarter Trillion Next Year and the four behemoths are on course to spend more than a quarter trillion dollars on AI infrastructure in the coming year.
Despite recent worries from Big Tech and others downstream on the sustainability of this AI investment, management teams have taken action to show that demand is still exceeding capacity and that AI revenue streams are close to and surpassing $10 billion.
As it rapidly approaches the quarter-trillion milestone, I examine the growth in AI investment from Big Tech this year below. Next week, I’ll go into detail about what this implies for the market’s largest benefit.
Since Microsoft, Meta, Alphabet, and Amazon were among the first to identify multi-billion dollar income streams from AI and generative AI solutions, AI Capex Accelerating Big Tech’s AI-fueled capital expenditures serve as a gauge for the larger AI industry. By allocating billions of dollars per quarter to AI infrastructure, the four are also spearheading the effort, indicating that they are still working to meet the need for AI and want to increase their investments in the field by 2025.
Let’s go back to 2023 to gain a better understanding of the trend of AI spending. At the start of that year, ChatGPT’s explosive growth laid the groundwork for AI’s swift rise to prominence.
Big Tech spent over $74 billion on capital expenditures in the first half of 2023. By the third quarter, that amount had increased to about $109 billion.
Big Tech spent around $104 billion in the first half of 2024, a 47% YoY increase. By Q3, that amount had increased to $170 billion, a 56% YoY increase.
Think about why Big Tech is investing billions in AI infrastructure worldwide to get a sense of why these four are driving spending this year and setting the stage for even greater spending in 2025. Why is Big Tech building up custom silicon or buying GPUs in bulk to provide cloud-based AI services to millions of business clients?
There are three parts to the answer:
1) According to a recent IDC projection, AI may have a trillion-dollar cumulative economic impact by 2030, with multi-trillion dollars expected to be impacted globally. The mobile economy, which gave rise to some of the current trillion-dollar IT behemoths, contributed roughly 7 trillion to the economy in 2023. Big Tech executives are aware of how critical it is to grasp and capitalize on an opportunity of this magnitude, and they will not let it slip through their fingers.
2) Since only Big Tech can afford the enormous computational capacity needed to construct larger models and double their sizes, the majority of genAI advancements are probably going to take place mostly in the clouds of the hyperscalers.
3) Three of the four companies claim that AI revenue is at least in the multibillion-dollar range, indicating that Big Tech is already reaping the benefits of AI. The long-term revenue potential could outstrip some of their current top revenue sources, given that there are millions to billions of customers for goods that can either be enhanced with AI integrations or targeted with AI features in subscriptions.
In May, I made this statement: The capital outlay expenditure (capex) of the information technology sector ‘may be in the region of 300 billion going on 500 billion dollars in aggregate for the year, with most of it towards AI infrastructure – which includes data center build or expansion, GPU buying and custom silicon work and more.
However, considering the rate of acceleration observed in Q3 and the predictions for Q4 and full-year spending, that number is probably already too low. Big Tech spent $64.9 billion in Q3, up 11% from the previous quarter. Amazon was the main driver of this growth, increasing capital expenditures by about $5 billion in a sequential manner.
Together, Microsoft and Amazon spent $42.6 billion on capital expenditures in the third quarter, while Alphabet continued to spend about $13 billion per quarter and Meta started to increase its expenditures.
Amazon indicated in Q3 that it will spend $75 billion on capital expenditures this year, while Meta tightened and increased its capital expenditure guidance to $38 billion to $40 billion. Together, the two companies anticipate spending close to $115 billion in 2024. The combined capital expenditures of the two will need to be close to $35 billion to reach that goal.
While Microsoft did not provide a specific capex forecast for this year, Alphabet anticipates Q4’s capex to be about in line with Q3’s as it continues to spend about $50 billion for the entire year. The two would spend almost $33 billion in Q4 if Microsoft’s spending remained constant from quarter to quarter.
Big Tech may spend an additional $70 billion in Q4, mostly for AI infrastructure, bringing their total capital expenditures for the year to about $240 billion, which is almost 15% more than what they had initially planned.
Don’t miss next week’s newsletter, as the I/O Fund will explain what this means for the largest beneficiary of this trend.
This AI-driven capital expenditure boom is expected to continue until 2025, as executives anticipate enduring demand for AI and the necessity to continue investing to meet demand and capture growth.
Executives Say AI Demand Is Here to Stay and Needs More Investment
This is a segment from the May newsletter titled Big Tech Q1 Earnings and I would like to paraphrase it: Historically, there has always been the relevance of capital expenditures on AI-related technologies to the investors, which has led to the increase in expenditure on AI Capital Expenditure. Given the “optimistic views of AI revenues in the billions and the high demand for GPUs where the supply cannot meet.” Understandably, the Major Companies in Technology are upping their spending by over 50% as opposed to 2023.
This was a recurring theme in Big Tech’s Q3 earnings calls. Examine the following statements made by executives:
Microsoft: To meet cloud demand, Microsoft spent about $10 billion on GPU and CPU servers in the most recent quarter. According to management, “demand continues to be higher than our available capacity.”
As stated by Chief Financial Officer Amy Hood, Microsoft prophesies that ‘capital expenditures will rise on a sequential basis given our cloud and AI demand signal. i.e. Also Microsoft,’ ‘reported investing in new cloud and AI infrastructure in Brazil, Italy, Mexico, and Sweden’ informed growth in its capacity in response to expected demand.
According to Hood, ‘Microsoft believes we will be able to do some supply-demand matching and so while we are talking about acceleration [in Azure] in the back half,’, especially about artificial intelligence, they ‘will get a good supply coming in over the second half of the year’.
The chairman of Amazon, Andy Jassy opined: “I think in the present scenario, there is hardly any business in which there is spare capacity, and there should be demand for that capacity. It is probably only in chips that supply could be put to use by companies in more than excess of what has already been supplied.”
He also said that AWS has, “more demand than we would be able to satisfy already had we an even bigger capacity today,” alluding to the significance of AWS services. “I believe the growth rate there might have a possibility of improving with time as we are looking at more and more growth in capacity,” he went on to explain further about the growth of AI technologies in AWS.
According to Jassy, Alphabet, Oracle, and other companies are also having trouble meeting demand because they are unable to buy enough GPUs or install enough bespoke accelerators in addition to GPUs. This means that AWS and Microsoft are not the only companies with supply constraints.
AWS’s need for fast-growing AI expenditures and long-term demand were both hinted at by Jassy. According to him, AI is “earlier stage [and] more fluid and dynamic than our non-AI part of AWS,” and “30,000 chips in a day” is not what customers are requesting. They are preparing ahead of time. We may therefore estimate how much we require based on very important demand signals.
Given that Amazon has dramatically increased its capital expenditures over the last two quarters, from $14.6 billion in Q1 to $22.6 billion in Q3, it’s intriguing that this comment was made. According to Jassy’s comment, AWS is experiencing far higher demand than they anticipated at the start of the year, which means they will need to invest significantly more in AI infrastructure, including data centers, servers, GPUs, and bespoke silicon.
Alphabet: While the company’s management had indicated an increase in expenses in 2025, the leader in search engines was somewhat more vague on the role AI would play in its cloud services. “Firstly, Developing capabilities in AI is not something easy, but it also has to the whole world, which we can do through our products and platforms “There is a necessity of having access to capital in adequate amounts.”, CFO Anat Ashkenazi acknowledged. Concerning other prospects of Alphabet, Ashkenazi stated, “We do see an increase coming in 2025 and we shall be more elaborate on that during the Q4 call, most probably a further increase than the same percentage step-up that we saw between 2023 and 2024.”.
Investments in AI strategies are expected to remain within the confines of a more conventional structure and as such I foresee a sustained amount of investment in building up such infrastructure because AI investments require somebody to put up an AI infrastructure – ‘somebody being the company. As Zuck referred to optimistic projections for the growth of the company, he expressed his disappointment that they were done in anticipation of limiting on AI to illnesses that already exist and should already provide high returns of investments throughout a couple of years.
CFO Susan Li went on to say, “Meta is increasing our infrastructure investments quite relatively this year and we are again expecting quite large growth in 2025.” In addition, she explained that in Q4, Meta is anticipating a significant jump across the above-mentioned factors partly also due to ‘uprisings in server spend and to a lower degree data center capex’ because of delivery and cash recognition dynamics.
AI Revenue Streams Emerging
AI revenue streams are fast emerging as Big Tech continues to splurge on AI, a sector where Microsoft is one of the pioneers since the software giant sees AI revenue soon hitting double digits.
Unfounded fears have been dispelled; these returns are on the rise. Moreover, it is heartening to note that we are only 2.5 years into this investment, and revenue from AI will cross a billion dollars yearly as soon as Q2,” remarked Microsoft Corporation’s chief executive officer, Satya Nadella. We expect this timeframe to be even shorter than any other business we have launched in at least the recent past.
Looking at it closer, AI accounted for 12 of the Azure growth in the most recent quarter, meaning that Azure’s AI run rate is already over $6 billion, while other growth comes from the Microsoft product suite: Power Platform having experienced a 4x YoY growth to over 600,000 users that were utilizing AI capabilities and 70% of Fortune 500 using Microsoft 365 Copilot.
Amazon typically does not reveal the percentage of revenue generation from products powered by AI. However, the company said: ‘The AI business within AWS is a multi-billion dollar revenue run rate juggernaut which is likely to record a triple digit year on year growth. And this particular stage of its evolution is growing more than 3 times faster than AWS itself grew.
La divisione AWS ha impiegato circa 2 anni per passare da quasi zero in vendite nel 2010, a vendite superiori al miliardo di dollari nel 2012 e altri 3 anni per quasi tre billion-in reconocimiento. The idea of AI developing at like a 3x increase at the multi-billion level at present already speaks volumes of the scope of the AI opportunity that is ahead and the appetite that there is in the market still unable to be satisfied.
Alphabet: A new update on AI revenue has not been provided by Alphabet. The last update from the second quarter was that “AI infrastructure and generative AI solutions for Cloud customers have already created billions in revenues and are used by more than 2 million developers.”
The management provided some additional aspects about accelerating the adoption of AI in all its offerings. “Gemini API Calls Have Increased By Almost 40 Times Within Six months, As Search ‘AI Overview’ Will Now Go Over 1 Billion Monthly Users.
Meta: In terms of clarity around how to monetize, Meta’s is the least clear because AI is primarily here to drive better ROI and conversions for advertisers, driving that advertising revenue higher.
According to the management, “Since the launch of Meta AI, the active monthly enhancements for AI-driven feed and video recommendations have reached over 500 million, which has resulted in a 6% engagement on Instagram this year alone and an 8% increase in Facebook’s growth time engagement.” Last month alone, over one million advertisers utilized our Gen AI features to design and publish over 15 million advertisements, and we predict that advertisers utilizing image-generating tools are experiencing as much as a 7 percent uptick in conversion rates, and there is much more where that came from.
What is not as evident, however, is what the monetization potential of these growths powered by AI would be, but optimistically management believes in the longer-term returns on investment that will result from putting money into AI strategies. Management elaborates, “Due to the availability of Meta AI, we have witnessed an upward trend in Facebook Time spent by 8% and 6% in Instagram engagement respectively this 2nd quarter alone, with more than 500 million active monthly improvements to the AI-powered feed and video recommendations on our platforms.”
Conclusion
The end spending by Big Tech on AI will only increase through the rest of 2024 and into 2025 as management teams continue to emphasize the need for more investment in this space to meet demand and complete the build-out of AI infrastructure. Microsoft leads the pack, with AI on the cusp of surpassing a $10 billion run rate, while Amazon and Alphabet see AI revenues in the billions. Here is more on AI
