Decades are happening in months.
A flurry of AI headlines piled up as infrastructure commitments climbed to the trillion-dollar range. In September alone, there were over $500B worth of contracts signed, totalling 40 GW of power. To put this into perspective, the US consumes about 450 GW of continuous power per year.
With the announcement of these deals came the chorus of calls for a bubble. It’s almost as though we now have a bubble of people calling for a bubble.
As blockbuster AI financing deals get announced, there’s increasing circularity amongst the key players in this ecosystem. This is creating a new chokepoint. It’s starting to look like OpenAI must succeed for this party to keep going.
Is this round-tripping? Or is this a virtuous flywheel?
Q3 Recap
Volatility Returns Amid AI Euphoria: Q3 brought some wild market swings, yet it ended with technology stocks soaring to new peaks. The S&P 500 jumped almost 8% during the quarter, while the tech-heavy Nasdaq posted double-digit gains, with both indices notching record highs. A widely anticipated Fed rate cut in September, surprisingly resilient earnings, and unrelenting enthusiasm for artificial intelligence helped defy the usual late-summer doldrums.
Under the surface, market leadership broadened beyond the Magnificent Seven mega-caps that dominated the first half: previously lagging giants like Tesla, Alphabet, and Apple surged ahead, while early-year leaders such as Microsoft and Meta took a breather. Small-cap and value stocks also grabbed a bigger share of gains as investors rotated into some less-expensive corners of the market.
At the same time, elevated valuations made traders more selective and unforgiving. Earnings surprises led to sharp dispersion in stock performance, and companies that disappointed felt the sting of idiosyncratic selloffs. But overall, Q3’s message was one of resilience: robust economic data and the AI boom’s second wind kept the bulls firmly in charge.
AI Infrastructure: Circular Deals and New Alliances
A wave of blockbuster partnerships in Q3 highlighted the circular nature of AI revenue. Big tech and AI firms are financing each other’s growth in ways that raise both excitement and eyebrows.
The centrepiece was NVIDIA’s major deal with OpenAI, announced in late September. In this arrangement, NVIDIA will invest up to $100B in OpenAI (via non-voting shares) and in return, OpenAI will use that cash to buy NVIDIA’s AI hardware: at least 10GW of cutting-edge NVIDIA systems.
In effect, NVIDIA is taking an equity stake in the world’s highest-profile AI startup and securing a long-term buyer for its chips. The move sent NVIDIA stock to new highs and underscored how tightly chipmakers and AI labs are now intertwined.
However, it also prompted debate over round-trip revenue. The concern here is that NVIDIA’s investment dollars come right back as OpenAI’s chip purchases, a self-reinforcing loop that is not traditional demand. Regulators took notice too, since such a deep alliance between the dominant AI chip supplier and a leading AI software firm raises antitrust questions about market fairness. NVIDIA maintains it will treat all cloud customers equally despite its OpenAI stake.
Microsoft’s AI bets also evolved in a circular fashion. Its multi-year investment in OpenAI, about $13B across rounds, was famously structured so that OpenAI paid much of that back to Microsoft via heavy use of Azure cloud services. Last quarter provided proof of how well that strategy is paying off. Azure’s revenue growth reaccelerated to 39% year-over-year, and its annualized run-rate approached the high tens of billions. Microsoft essentially funded OpenAI’s research war chest on the condition that OpenAI spend a chunk of it on Azure, boosting Microsoft’s own cloud business. Wall Street has begun to question how much of the big cloud providers’ AI revenue comes from these kinds of captive deals.
Microsoft, for its part, says its accounting is conservative and has also begun to diversify. Late in Q3, it announced that Anthropic’s AI models will be integrated into Microsoft 365 Copilot alongside OpenAI’s, part of a push to reduce dependence on a single model vendor. Even as Microsoft remains OpenAI’s key backer and cloud host, it’s hedging its bets by embracing an OpenAI rival. This multi-model approach, plus Microsoft’s work on its own AI chips and models, shows how complex and co-opetitive the AI ecosystem has become.
Meanwhile, OpenAI itself is aiming to become a hyperscaler. The company and its allies have embarked on a globe-spanning infrastructure binge, raising staggering sums to build out AI supercomputing capacity. OpenAI’s leadership has made it clear they intend to make a very aggressive infrastructure bet now to meet future AI needs. NVIDIA CEO Jensen Huang framed its direct sales to OpenAI as preparing them for the day when OpenAI runs its own data centres as a self-hosted hyperscaler.
Each 1-GW mega-data centre could cost tens of billions all-in, but that hasn’t stopped OpenAI from signing deal after deal. In the U.S., OpenAI partnered with Oracle and others on Stargate, a plan to invest roughly $500B building out at least 10 GW of AI cloud capacity across multiple new data centres. It inked a separate multi-hundred-billion-dollar cloud contract with Oracle for U.K. and European expansion. Add in the 10 GW from Nvidia’s deal and about 6 GW from a new AMD deal, plus other regional projects, and OpenAI has signed commitments approaching $1T of infrastructure in 2025 alone.
This planned capacity far exceeds current needs. It’s a bet that future demand for AI will mushroom dramatically. Critics have called it an AI “build it and they will come” strategy. The counter from OpenAI is that everyone in the industry needs to chip in, from electrons to model distribution, and that future models will create unprecedented economic value. In this view, these massive bets will eventually pay for themselves, even if today’s revenue is a far cry from the trillion-dollar scale of the investments.
New Entrants: AMD and xAI Strike Back
Until recently, NVIDIA had a virtual lock on the AI compute market, but Q3 saw major moves that broadened the playing field. AMD secured a landmark chip supply deal with OpenAI in early October. The multi-year agreement commits OpenAI to use hundreds of thousands of AMD GPUs starting in 2026, based on AMD’s forthcoming Instinct MI450 accelerators. In return, OpenAI received warrants to buy up to a 10% equity stake in AMD at $0.01 per share if certain deployment and performance milestones are met.
This aligns incentives to ramp usage. For AMD, the upside is enormous. Management outlined a path to more than $100B in incremental revenue over the next four years tied to OpenAI and related customers. The market reaction was immediate. AMD shares spiked by more than 30% in a single session, with analysts calling the deal a major vote of confidence in AMD’s AI roadmap and software stack.
Even so, NVIDIA’s throne is safe for now. The market leader still sells every chip it can produce and will remain OpenAI’s primary supplier in the near term. OpenAI emphasized that the AMD deal is incremental and that it will continue buying more NVIDIA systems in parallel. Translation: OpenAI wants it all. The world needs much more compute.
Not to be outdone, Elon Musk’s xAI also entered the fray with a NVIDIA-powered twist. In Q3, xAI was revealed to be raising a colossal financing package to build out its AI computing platform. NVIDIA is playing a central role: the structure involves a mix of equity and debt channelled into a vehicle that will purchase NVIDIA GPUs in bulk and lease them to xAI. NVIDIA is also participating as an equity investor.
Much like the OpenAI arrangement, NVIDIA is helping to finance its own customer. xAI gets the chips and priority access to supply. NVIDIA secures another giga-scale buyer. The chips are destined for Colossus, xAI’s data centre in Nevada, which Musk aims to grow toward hundreds of thousands of GPUs. The setup effectively finances purchases of NVIDIA’s own hardware while guaranteeing xAI a GPU supply lifeline in a tight market. Observers quickly noticed the familiar pattern of circular financing. Musk’s ambition is to assemble a closed-loop AI ecosystem that can tap across his companies.
With NVIDIA’s help and a lot of capital, xAI now has a shot to build a hyperscale AI cluster from scratch. The risk is obvious. xAI is pre-spending billions on capacity without proven product demand yet, a microcosm of the entire industry’s leap of faith.
Web of Interdependence and CapEx Supercycle
Taken together, these developments create an increasingly interdependent web between the hyperscalers, the AI model labs, and the semiconductor makers. Cloud giants invest in AI startups and guarantee them compute resources. Those startups lock in long-term orders for tens of billions of dollars of chips and cloud capacity, often from their very investors. Chipmakers take equity stakes in their biggest customers and sometimes fund those customers’ purchases, effectively pulling demand forward. The result is a self-reinforcing loop fueling the current AI boom.
Enormous capital expenditures are being poured into AI infrastructure on the expectation of even more enormous future revenues. This has created an AI CapEx supercycle. Hyperscalers’ data centre spending forecasts have been revised sharply upward for 2024 through 2026, and may still be too low. Every time companies revisit AI capacity plans, they end up raising the targets yet again. Earnings commentary has reinforced this as several big tech CEOs emphasized they would rather overspend on AI capacity buildout than get left behind, and they’re following through.
Crucially, monetization is finally catching up. After years of “pre-revenue” hand-wringing, we now have tangible signs of AI-driven revenue growth at scale. Azure’s massive cloud business is growing nearly 40% in part due to AI services. OpenAI has roughly doubled to an annualized revenue run-rate in the tens of billions within months.
It’s difficult to overstate how parabolic that trajectory is for a company that had near-zero revenue three years ago. Adoption is being driven by viral consumer usage and rapidly maturing enterprise willingness to pay for AI capabilities that barely existed not long ago. In other words, the “R” in ROI for AI is finally showing up, and at large numbers. That momentum is encouraging even more investment, creating a virtuous cycle for those betting on AI.
The big question is how long this virtuous, or circular, cycle can sustain itself. These partnerships clearly obscure visibility into what end-user demand truly is. Are we witnessing organic pull-forward demand for AI everywhere, or is a significant portion of today’s demand generated by these partnerships where vendors buy from each other? If everyone is buying growth via round-trip deals, the underlying demand might be lower than it appears once the music stops. The business case for all the new AI capacity remains unproven over the long run. If lofty revenue projections do not materialize, the industry could be left with overbuilt data centres and a hangover of excess.
On the other hand, we're likely still in the very early innings of AI adoption. Soon, every company and consumer could be using AI-powered services daily, justifying the infrastructure. From that perspective, these alliances are simply bridging the financing gap until AI’s killer apps and broader usage explode.
For now, Q3 2025 shows the stock market is willing to bet on the optimistic scenario. Tech investors pushed valuations higher across the board, effectively endorsing the massive spending as prudent and necessary. The AI hype cycle has become a flywheel. Early revenue wins are bolstering the narrative that AI will be transformative, which in turn attracts more capital to fuel even bigger bets. It’s a feedback loop of hype, money, and increasingly, real product traction. We will still need more compute, which means more infrastructure, which means more power.
Don't expect it to slow yet.
We believe we’re still just getting started.
Wrapping Up
Q3 brought record highs for tech on the back of AI’s promise, while also spotlighting an ecosystem that is financially circular. Interdependent partnerships between chipmakers, cloud providers, and AI labs are sustaining an extraordinary boom, even as they raise concerns about round-trip revenue and opaque demand signals.
Enjoy the gains, but keep a close eye on fundamentals. So far, we believe AI is delivering growth as advertised. As long as that continues, the circular flow of capital into AI will likely keep markets buoyant. The message to companies and investors remains the same: get on board with AI’s platform shift, or risk being left behind.
Strong Convictions. Loosely Held.
- Nicholas Mersch, CFA
Sources: Charts are sourced to Bloomberg L. P.
The content of this document is for informational purposes only and is not being provided in the context of an offering of any securities described herein, nor is it a recommendation or solicitation to buy, hold or sell any security. The information is not investment advice, nor is it tailored to the needs or circumstances of any investor. Information contained in this document is not, and under no circumstances is it to be construed as, an offering memorandum, prospectus, advertisement or public offering of securities. No securities commission or similar regulatory authority has reviewed this document, and any representation to the contrary is an offence. Information contained in this document is believed to be accurate and reliable; however, we cannot guarantee that it is complete or current at all times. The information provided is subject to change without notice.
Commissions, trailing commissions, management fees and expenses all may be associated with investment funds. Please read the prospectus before investing. If the securities are purchased or sold on a stock exchange, you may pay more or receive less than the current net asset value. Investment funds are not guaranteed, their values change frequently, and past performance may not be repeated. Certain statements in this document are forward-looking.
Forward-looking statements (“FLS”) are statements that are predictive in nature, depend on or refer to future events or conditions, or that include words such as “may,” “will,” “should,” “could,” “expect,” “anticipate,” intend,” “plan,” “believe,” “estimate” or other similar expressions. Statements that look forward in time or include anything other than historical information are subject to risks and uncertainties, and actual results, actions or events could differ materially from those set forth in the FLS. FLS are not guarantees of future performance and are, by their nature, based on numerous assumptions. Although the FLS contained in this document are based upon what Purpose Investments and the portfolio manager believe to be reasonable assumptions, Purpose Investments and the portfolio manager cannot assure that actual results will be consistent with these FLS. The reader is cautioned to consider the FLS carefully and not to place undue reliance on the FLS. Unless required by applicable law, it is not undertaken, and specifically disclaimed, that there is any intention or obligation to update or revise FLS, whether as a result of new information, future events or otherwise.