Artificial intelligence has now marked a full 70 years since its birth at the Dartmouth Summer Workshop in 1956. In his latest research report, Deutsche Bank Research thematic strategist Adrian Cox systematically traces AI’s complete evolutionary arc\u2014from symbolic logic to large language models\u2014distilling 14 key insights that provide investors with a historical reference framework for assessing the sustainability of the current AI boom. The report’s core thesis: the current wave of AI investment and valuation expansion is replaying a pattern seen repeatedly across history’s technological revolutions, and understanding “context” is the key to grasping where AI goes from here.<\/p>\n
The report opens by noting that the defining characteristic of AI progress is its nonlinearity. Plotting historical training compute data on a logarithmic scale reveals that since 1956, the compute used to train major AI systems has grown across dozens of orders of magnitude\u2014a trend almost entirely obscured by linear charting. Exponential growth is intuitively easy to underestimate, and this represents the first cognitive hurdle in understanding the AI wave. Closely related is the fact that AI’s pace of advancement has already surpassed Moore’s Law. Traditional computing power doubles every 18 to 24 months, but since the advent of deep learning, compute has grown at roughly 4x per year\u2014far exceeding the approximately 1.4x annual growth rate seen before the deep learning era. The report attributes this to the compounding effect of simultaneous improvements across multiple dimensions: larger system scale, enhanced memory, and algorithmic optimization.<\/p>\n
From a technological lineage perspective, AI has cycled through multiple generations over seven decades: symbolic logic, expert systems, statistical machine learning, deep learning, and now large language models. Each dominant technology has experienced its own rise-and-fall cycle, with some approaches superseded while others continue evolving in parallel. The report notes that large language models may eventually yield to new paradigms such as “world models”\u2014technological generational shifts do not bend to the will of current incumbents. Historical market share turnover reinforces this point: Internet Explorer once crushed Netscape, only to be displaced by Chrome. In today’s generative AI platform competition, ChatGPT leads in monthly visits, but Google Gemini, DeepSeek, and Claude are all rapidly closing the gap. Early advantage does not equal a lasting moat.<\/p>\n
The sudden emergence of Chinese AI models\u2014DeepSeek being the prime example\u2014may appear to be an overnight success, but it is in fact the product of years of accumulated R&D investment. Data shows that China’s total R&D spending surpassed that of the US in 2024, and its catch-up speed in the number of major AI models is equally striking. In terms of AI patent grants, China’s growth curve also leads other economies by a wide margin. The report emphasizes that for investors, competitive landscape shifts often accumulate beneath the surface for years before becoming visible.<\/p>\n
The Cost Paradox and Hardware Bottlenecks<\/p>\n
The report invokes the Jevons Paradox to illustrate a key dynamic: dramatic declines in AI usage costs do not reduce aggregate spending\u2014they stimulate surging demand. Since 2006, GPU compute costs have fallen by more than 99%, yet according to International Energy Agency (IEA) projections, global data center electricity consumption will double from 2024 to 2030. Lower marginal costs mean more application scenarios and higher total demand.<\/p>\n
Every technological revolution has depended on massive hardware investment, and AI is no exception. Since ChatGPT’s launch in November 2022, data center, hardware, and chip-related sectors within the Russell 1000 index have delivered total returns far exceeding those of the software sector. Unlike historical constraints tied to consumer device adoption, the current bottleneck lies primarily on the supply side of AI chips. Hourly rental prices for Nvidia’s H100 GPU chips have fluctuated persistently over recent quarters, reflecting structural supply-demand tension.<\/p>\n
As the AI boom intensifies, US semiconductor imports have surged dramatically. Meanwhile, global supply chain concentration continues to rise, with products becoming increasingly dependent on single sources. The report notes that while globalization has made technology cheaper, it has also exposed supply chains to heightened geopolitical and concentration risks.<\/p>\n
The Productivity J-Curve and Valuation Warnings<\/p>\n
Technological revolutions typically exhibit a productivity “J-curve” effect: costs materialize before benefits. The report cites data showing that fewer than half of US employees currently use AI in their work, dedicating an average of 6% of work time to AI-related tasks while saving only about 2% of work hours. Consumer-side adoption has set historical records\u2014generative AI has spread faster than the internet or personal computers. However, directly monetizable enterprise applications are advancing considerably more slowly: downloading an app takes minutes, while restructuring corporate operations around new technology takes years. AI adoption rates are highest in information, professional services, and finance\/insurance sectors, while manufacturing lags behind.<\/p>\n
On the Shiller cyclically adjusted price-to-earnings (CAPE) basis, the S&P 500’s current valuation is now comparable to the 2000 dot-com bubble peak, sitting in the highest range in 150 years. Historically, nearly every valuation peak has corresponded to a transformative technology wave. The report catalogs the timeline of valuation peaks across successive technology waves:<\/p>\n
YearTechnology Wave1899Electrification1929Radio and Automobiles1966Electronics2000Internet2026Artificial Intelligence<\/p>\n
On the revenue front, OpenAI and Anthropic’s current growth rates have already exceeded the historical peak levels of comparable companies from earlier eras. But the report simultaneously cautions that to justify the forward revenue expectations embedded in current valuations, the two companies’ growth trajectories would need to be “even more exceptional” relative to the historical paths of tech giants like Google, Meta, and Nvidia.<\/p>\n
The report concludes with the long-term trend of S&P 500 earnings per share, showing that the metric has grown at roughly 6.5% annually since 1935, maintaining trend stability through multiple wars and recessions. At the GDP level, the report outlines three AI scenarios: a baseline trend, an AI-driven moderate acceleration (approximately 2.1% average annual growth over the next decade), and two extremes under a “singularity” scenario\u2014ranging from technological utopia to human extinction. The report maintains a neutral stance on these outcomes but reminds investors: whether current market pricing aligns more closely with historical trends or extreme scenarios is the central question worth continuously tracking.<\/p>\n","protected":false},"excerpt":{"rendered":"Artificial intelligence has now marked a full 70 years since its birth at the Dartmouth Summer Workshop in…\n","protected":false},"author":2,"featured_media":74215,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[21213],"tags":[55040,23163,698,55041,21214,2869,55042,15228,22553,48779],"class_list":["post-74214","post","type-post","status-publish","format-standard","has-post-thumbnail","category-deutsche-bank","tag-adrian-cox","tag-anthropic","tag-chatgpt","tag-deepseek","tag-deutsche-bank","tag-international-energy-agency","tag-jevons-paradox","tag-nvidia","tag-openai","tag-sp-500"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/germany\/wp-json\/wp\/v2\/posts\/74214","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/germany\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/germany\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/germany\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/germany\/wp-json\/wp\/v2\/comments?post=74214"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/germany\/wp-json\/wp\/v2\/posts\/74214\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/germany\/wp-json\/wp\/v2\/media\/74215"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/germany\/wp-json\/wp\/v2\/media?parent=74214"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/germany\/wp-json\/wp\/v2\/categories?post=74214"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/germany\/wp-json\/wp\/v2\/tags?post=74214"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}