Wall Street is buzzing with concern about the revenue numbers coming out of big tech’s cloud divisions. With massive investments in AI startups and related projects, there’s a growing suspicion that growth is being artificially inflated. The crux of the issue revolves around a practice known as “revenue round tripping.” This occurs when a large tech company invests in an AI startup, which then turns around and spends that money on the tech company’s cloud and AI services. Essentially, it appears as though these tech giants are buying their own growth.
Consider the example of Amazon Web Services (AWS). Recently, AWS made a hefty $4 billion investment in Anthropic, an AI startup. Conveniently, Anthropic agreed to use AWS as its primary cloud provider. Google and Microsoft have engaged in similar practices. Google invested in AI ventures which later utilized Google Cloud Platform (GCP) for their operations. Microsoft, on the other hand, invested in OpenAI, and surprise, OpenAI has been using Microsoft’s Azure as its cloud infrastructure. Oracle joined the fray by partnering with Cohere shortly after investing in the startup.
Analysts and investors are raising eyebrows over these deals as they could potentially skew cloud revenue figures. Business Insider brought these concerns to the forefront last year, highlighting how these revenue round-tripping arrangements might inflate cloud revenue numbers artificially. One analyst, Jaluria, expressed concerns that AWS revenue might include costs associated with training Anthropic models. Similarly, questions are being raised about GCP revenue from internally training Gemini models or Oracle Cloud Infrastructure (OCI) revenue from training Cohere models.
The primary issue here is the comparability between different cloud vendors. If revenue figures are bolstered by these round-tripping deals, it challenges the narrative of a broad-based recovery in cloud workloads. This is particularly pertinent in an era where cloud spending growth has slowed. Many businesses are tightening their belts amid a sluggish economy and high inflation. If the apparent recovery in cloud services is largely due to these internal transactions, then the optimism surrounding the cloud sector might be misplaced.
Interestingly, Microsoft might be an exception. According to RBC, Microsoft disclosed that it does not account for any revenue from OpenAI training its GPT models on Azure’s infrastructure. This transparency might give Microsoft a slight edge in the eyes of scrutinizing investors. On the other hand, Amazon has remained tight-lipped. An Amazon spokesperson declined to clarify whether AWS revenue figures include cloud spending by Anthropic.
As the cloud landscape continues to evolve, these revenue round-tripping practices will likely come under greater scrutiny. Investors are becoming increasingly aware and skeptical of the true sources behind the revenue figures. Transparency and clear disclosures will be paramount in maintaining investor confidence. The next time you hear about a cloud revenue surge, remember to take a closer look – it might just be a case of money going around in circles.