Artificial intelligence investment and accounting auditing have become the focus... from Meta to AWS, all have launched "profitability verification."

As the wave of generative artificial intelligence sweeps across the entire industry, issues related to corporate accounting practices have resurfaced. As investors begin to scrutinize whether the massive investments in AI-related expenditures are truly translating into actual returns, the accounting treatment of AI infrastructure, from GPU depreciation to data center accounting methods, has become a hot topic of discussion.

According to Dave Vellante from Breaking Analysis, recent analyses suggest that the GPUs used in AI factories may have a longer lifespan in terms of durability and performance than the current depreciation standards. However, Meta's accounting practices remain a point of contention. According to some reports, Meta has billions of dollars in data center investments that have not been recorded on the books, which further exacerbates market concerns over the uncertainty of its AI investment returns.

Doubts regarding AI technology itself are also spreading. AI expert Ilya Sutskever, who has long pointed out the limitations of deep learning large-scale model expansion strategies, claimed that “the era of scale expansion has come to an end, and the future will be dominated by research-driven approaches in the field,” signaling the arrival of a research-driven AI era. This has also sparked debates about the actual utility of data center infrastructure that invests massive amounts of electricity and capital.

However, the technology frontier remains active. OpenAI has launched a new shopping research tool to enhance personalized product search capabilities, while Anthropic has released its flagship model Claude Opus 4.5 for continuous iteration and upgrading. Microsoft has also introduced the low-power PC dedicated AI model “Para-7B” to maintain its momentum. Meanwhile, Jeff Bezos' secret AI company “Prometheus Project” has recently been reported to acquire the startup “General Intelligence.”

Corporate investment momentum remains strong. HarmonicAI raised $120 million to enhance its mathematical reasoning capabilities, while financial AI startup ModelML also secured $75 million. Amazon Web Services' decision to implement a super investment plan worth $500 billion to strengthen the AI supercomputing capabilities of the U.S. government has attracted market attention.

However, hardware and server manufacturers that are more sensitive to economic conditions have delivered mixed results. Dell and NetApp benefited from AI demand and achieved performance growth, while HPE cooled investment sentiment due to weak performance guidance and large-scale layoff plans. Nutanix and Workday also failed to avoid stock price declines due to disappointing performance outlooks.

The upcoming AWS annual event “re:Invent” is expected to focus on the sustainability of generative AI business models, aside from the current AI accounting controversy. Informal observations suggest that Amazon may collaborate with OpenAI to enhance the relatively lagging competitiveness of generative AI.

With the wave of the AI era surpassing the level of technological achievements and spreading to the field of accounting regulation, companies are entering a new era where they must not only be judged by their technological strength but also by their financial transparency and long-term profit strategies.