Latest / USA Daily News / Ken Griffin Warns AI Is “Almost Certainly” in a Bubble — Is Wall Street at Risk

Transcript

Ken Griffin Warns AI Is “Almost Certainly” in a Bubble — Is Wall Street at Risk · USA Daily News

• 0:00
  Speaker 1: The artificial intelligence revolution has become one of the most
• 0:04
  Speaker 1: powerful forces shaping global financial markets, attracting extraordinary investments from
• 0:10
  Speaker 1: some of the world's largest technology companies, financial institutions, and
• 0:16
  Speaker 1: private investors. Billions of dollars are flowing into advanced computer chips,
• 0:21
  Speaker 1: data centers, cloud infrastructure, and increasingly sophisticated AI models. As
• 0:27
  Speaker 1: businesses race to secure a position in what many believe
• 0:30
  Speaker 1: will be the next major transformation of the global economy.
• 0:35
  Speaker 1: Yet beneath this excitement, an uncomfortable question is beginning to
• 0:39
  Speaker 1: dominate conversations across Wall Street. What happens if investors have
• 0:44
  Speaker 1: become too optimistic about how quickly artificial intelligence will generate profits?
• 0:49
  Speaker 1: That question has gained renewed attention following a warning from
• 0:53
  Speaker 1: Ken Griffin. the billionaire founder and chief executive of Citadel,
• 0:58
  Speaker 1: one of the world's most prominent investment firms. Speaking at
• 1:02
  Speaker 1: Citadel's Future of Global Markets conference in New York on
• 1:05
  Speaker 1: October 20, 2026, Griffin said investors are almost certainly in
• 1:11
  Speaker 1: the middle of an artificial intelligence bubble, raising concerns about
• 1:15
  Speaker 1: whether the enormous sums being committed to the technology will
• 1:19
  Speaker 1: ultimately produce returns capable of justifying current expectations. Griffin's assessment
• 1:25
  Speaker 1: carries considerable weight because of his position within the financial
• 1:28
  Speaker 1: industry and his experience navigating major market cycles. Citadel has
• 1:34
  Speaker 1: built its reputation through sophisticated investment strategies, while Griffin has
• 1:39
  Speaker 1: spent decades observing how investor psychology, technological change, and the
• 1:44
  Speaker 1: movement of capital can combine to create extraordinary opportunities as
• 1:49
  Speaker 1: well as significant risks. His warning does not necessarily mean
• 1:53
  Speaker 1: that artificial intelligence is destined to fail or that every
• 1:58
  Speaker 1: company connected to the technology is overvalued. Instead, it highlights
• 2:03
  Speaker 1: a recurring problem in financial markets. Investors can correctly identify
• 2:08
  Speaker 1: a revolutionary technology while simultaneously paying too much for the
• 2:13
  Speaker 1: businesses expected to benefit from it. When expectations rise faster
• 2:17
  Speaker 1: than actual earnings, Even companies with promising products and strong
• 2:22
  Speaker 1: long-term prospects can become vulnerable to sharp declines in their
• 2:26
  Speaker 1: share prices. The distinction between technological progress and investment performance
• 2:32
  Speaker 1: is therefore central to understanding Griffin's concerns. The current AI
• 2:36
  Speaker 1: boom has developed around the belief that advanced computing systems
• 2:40
  Speaker 1: will fundamentally change how businesses operate, how workers perform their jobs,
• 2:46
  Speaker 1: and how products and services are developed. Generative AI tools
• 2:50
  Speaker 1: can produce text, analyze information, write software, assist with research,
• 2:55
  Speaker 1: and automate tasks that previously required substantial human effort. As
• 3:00
  Speaker 1: these capabilities improve, investors have begun imagining a future in
• 3:05
  Speaker 1: which companies operate more efficiently, develop products faster, and generate
• 3:10
  Speaker 1: greater profits with fewer resources. Technology firms have responded by
• 3:15
  Speaker 1: expanding their AI operations, purchasing specialized processors, constructing data centers,
• 3:22
  Speaker 1: and investing heavily in the computing infrastructure required to train
• 3:26
  Speaker 1: and operate increasingly capable models. For investors, these developments have
• 3:32
  Speaker 1: created a compelling growth narrative. Companies positioned at important points
• 3:37
  Speaker 1: in the AI supply chain appear capable of benefiting from
• 3:40
  Speaker 1: demand that could continue for years. And the prospect of
• 3:43
  Speaker 1: capturing a large share of this emerging market has encouraged
• 3:47
  Speaker 1: investors to accept valuations that would have seemed difficult to
• 3:51
  Speaker 1: justify under more conventional assumptions. However, the financial markets are
• 3:56
  Speaker 1: not simply valuing the technology that exists today. They are
• 3:59
  Speaker 1: also assigning enormous importance to expectations about what AI might
• 4:04
  Speaker 1: accomplish in the future. This creates a difficult challenge because
• 4:08
  Speaker 1: the commercial value of a transformative technology often becomes clear
• 4:12
  Speaker 1: only after years of experimentation, competition, and investment. Businesses may
• 4:18
  Speaker 1: spend heavily on AI systems before understanding which applications will
• 4:23
  Speaker 1: produce reliable financial benefits. Some companies may discover that automated
• 4:28
  Speaker 1: tools significantly improve productivity, while others may struggle to integrate
• 4:32
  Speaker 1: the technology into their operations or persuade customers to pay
• 4:36
  Speaker 1: enough to cover the associated costs. Even when AI delivers
• 4:41
  Speaker 1: measurable benefits, those gains do not automatically translate into higher
• 4:45
  Speaker 1: profits for every company involved. Competition can force prices downward.
• 4:51
  Speaker 1: Expensive infrastructure can absorb much of the revenue generated and
• 4:55
  Speaker 1: customers may demand lower costs as AI tools become more
• 4:59
  Speaker 1: widely available. Investors who assume that widespread adoption will guarantee
• 5:04
  Speaker 1: exceptional financial returns could therefore be overlooking the complicated process
• 5:09
  Speaker 1: through which technological innovation becomes sustainable commercial success. The debate
• 5:14
  Speaker 1: over whether artificial intelligence has entered a financial bubble cannot
• 5:19
  Speaker 1: be understood without examining how the technology became one of
• 5:23
  Speaker 1: the largest investment opportunities in modern financial history. What began
• 5:28
  Speaker 1: as a series of advances in machine learning, computing power,
• 5:32
  Speaker 1: and automated data processing has evolved into a global race
• 5:35
  Speaker 1: involving technology giants, semiconductor manufacturers, cloud computing providers, startups, governments,
• 5:44
  Speaker 1: and some of the world's most powerful investment funds. The
• 5:47
  Speaker 1: promise of AI is no longer limited to producing more
• 5:51
  Speaker 1: accurate predictions or automating simple tasks. Modern systems can generate
• 5:56
  Speaker 1: convincing text and images, assist with software development, analyze enormous
• 6:01
  Speaker 1: quantities of information, and support complex business operations. As these
• 6:07
  Speaker 1: capabilities have expanded, Investors have increasingly viewed artificial intelligence as
• 6:13
  Speaker 1: a foundational technology that could reshape industries in much the
• 6:17
  Speaker 1: same way electricity, personal computers, and the internet changed the
• 6:21
  Speaker 1: global economy. This belief has encouraged companies to spend aggressively,
• 6:27
  Speaker 1: while financial markets have rewarded businesses perceived to be positioned
• 6:31
  Speaker 1: at the center of the transformation. The investment boom accelerated
• 6:35
  Speaker 1: as generative AI became more visible to the public and
• 6:39
  Speaker 1: businesses began exploring practical uses for increasingly capable models. Companies
• 6:45
  Speaker 1: that once treated AI as a specialized research project started
• 6:49
  Speaker 1: integrating it into customer service, marketing, cybersecurity, product development, financial analysis,
• 6:57
  Speaker 1: and internal administration. Executives began considering how intelligent software could
• 7:02
  Speaker 1: reduce repetitive work. improve decision-making, and allow employees to complete
• 7:07
  Speaker 1: tasks more quickly. Investors responded by searching for companies that
• 7:12
  Speaker 1: could supply the tools and infrastructure required to support this transition.
• 7:17
  Speaker 1: The opportunity appeared enormous because AI could potentially become part
• 7:21
  Speaker 1: of almost every sector of the economy, from healthcare and
• 7:25
  Speaker 1: manufacturing to banking, education, entertainment, and transportation. Unlike a technology
• 7:32
  Speaker 1: designed for a single market, artificial intelligence promised to create
• 7:37
  Speaker 1: multiple revenue streams across industries, making it particularly attractive to
• 7:42
  Speaker 1: investors seeking long-term growth. At the center of this expansion
• 7:46
  Speaker 1: are semiconductor companies producing the advanced processors needed to train
• 7:51
  Speaker 1: and operate sophisticated AI models. These processors perform the enormous
• 7:56
  Speaker 1: number of mathematical calculations required by modern artificial intelligence systems,
• 8:02
  Speaker 1: and demand for them has increased as technology companies compete
• 8:05
  Speaker 1: to develop more capable products. Specialized chips have consequently become
• 8:10
  Speaker 1: a strategic resource, with manufacturers and their suppliers benefiting from
• 8:15
  Speaker 1: the urgency surrounding AI development. Yet the significance of semiconductor
• 8:20
  Speaker 1: companies extends beyond the sale of individual components. Their products
• 8:25
  Speaker 1: sit within a broader ecosystem that includes advanced manufacturing equipment, memory,
• 8:31
  Speaker 1: networking systems, cooling technology, electricity infrastructure, and specialized software. Every
• 8:38
  Speaker 1: expansion in AI computing capacity creates demand across several connected industries,
• 8:44
  Speaker 1: encouraging investors to view the entire supply chain as a
• 8:47
  Speaker 1: potential beneficiary of sustained technological growth. Cloud computing providers have
• 8:53
  Speaker 1: also become central to the investment story. Training and operating
• 8:58
  Speaker 1: large AI models requires access to extensive computing resources that
• 9:03
  Speaker 1: many businesses cannot economically maintain on their own. Cloud platforms
• 9:08
  Speaker 1: allow customers to purchase computing capacity as needed. reducing the
• 9:12
  Speaker 1: need to build their own facilities and purchase all the
• 9:15
  Speaker 1: necessary equipment. Major technology companies have therefore committed substantial resources
• 9:21
  Speaker 1: to expanding their data centers, acquiring processors, improving networking infrastructure,
• 9:27
  Speaker 1: and securing electricity supplies. Their strategy rests on the expectation
• 9:32
  Speaker 1: that AI developers and corporate customers will require more computing
• 9:37
  Speaker 1: capacity as models become increasingly capable and applications spread throughout
• 9:42
  Speaker 1: the economy. If that demand continues growing, cloud providers could
• 9:47
  Speaker 1: generate significant revenue from infrastructure that supports thousands of different customers. However,
• 9:54
  Speaker 1: the scale of these commitments also means that investors must
• 9:57
  Speaker 1: determine whether future demand will be large enough to justify
• 10:01
  Speaker 1: the expense of building and maintaining such facilities. The competition
• 10:05
  Speaker 1: among technology companies has created an additional force driving investment.
• 10:11
  Speaker 1: No major platform wants to risk falling behind rivals in
• 10:14
  Speaker 1: a market that could influence future productivity and consumer services,
• 10:19
  Speaker 1: and corporate profitability. Companies are therefore investing not only to
• 10:24
  Speaker 1: meet existing demand, but also to preserve their strategic position.
• 10:28
  Speaker 1: Behind the extraordinary growth of artificial intelligence lies a financial
• 10:32
  Speaker 1: challenge that could determine whether the current investment boom becomes
• 10:36
  Speaker 1: a lasting economic success or one of the most expensive
• 10:40
  Speaker 1: miscalculations in modern technology history. Developing advanced AI systems requires
• 10:46
  Speaker 1: far more than talented researchers and powerful software. Companies must
• 10:51
  Speaker 1: secure specialized processors, construct and operate data centers, purchase networking equipment,
• 10:58
  Speaker 1: obtain reliable electricity, and continually improve the infrastructure needed to
• 11:03
  Speaker 1: train and run increasingly sophisticated models. These requirements create enormous
• 11:08
  Speaker 1: upfront costs. often years before the businesses making the investments
• 11:13
  Speaker 1: can demonstrate that the resulting products will generate sufficient revenue.
• 11:17
  Speaker 1: For investors already concerned about the possibility of an AI bubble,
• 11:22
  Speaker 1: this mismatch between immediate spending and uncertain future returns has
• 11:27
  Speaker 1: become one of the most important risks facing the industry.
• 11:30
  Speaker 1: The financial commitments involved are particularly significant because AI infrastructure
• 11:35
  Speaker 1: is expensive to build and difficult to adjust quickly when
• 11:39
  Speaker 1: market conditions change. A technology company can revise a software
• 11:43
  Speaker 1: product or reduce certain operating expenses relatively rapidly, but constructing
• 11:49
  Speaker 1: a large data center requires land, buildings, electrical systems, cooling equipment,
• 11:55
  Speaker 1: advanced networking, and access to suitable computing hardware. These projects
• 12:00
  Speaker 1: also depend on permits, construction schedules, utility connections, and long-term
• 12:06
  Speaker 1: infrastructure planning. Once substantial capital has been committed, management may
• 12:11
  Speaker 1: have limited flexibility to reverse course without absorbing considerable losses.
• 12:16
  Speaker 1: Companies therefore make investment decisions based on forecasts about future
• 12:21
  Speaker 1: computing demand, anticipated customer growth, and the expected commercial importance
• 12:26
  Speaker 1: of AI. When those forecasts prove accurate, early investments can
• 12:31
  Speaker 1: establish valuable competitive advantages. When they are too optimistic, businesses
• 12:37
  Speaker 1: may become burdened by expensive assets that generate less revenue
• 12:41
  Speaker 1: than originally expected. One of the central questions is whether
• 12:45
  Speaker 1: the revenue generated by AI services will eventually justify the
• 12:49
  Speaker 1: cost of producing them. Training advanced models can require extensive
• 12:53
  Speaker 1: computing resources, while operating those systems for millions of users
• 12:58
  Speaker 1: creates continuing expenses. Every request processed by an AI system
• 13:03
  Speaker 1: consumes computing capacity, electricity, and infrastructure resources, although the exact
• 13:09
  Speaker 1: cost varies according to the model, hardware, workload, and operating efficiency.
• 13:15
  Speaker 1: Companies must balance these expenses against the prices customers are
• 13:19
  Speaker 1: willing to pay. Subscription services may attract large numbers of
• 13:23
  Speaker 1: users but struggle to generate sufficient revenue if many customers
• 13:28
  Speaker 1: choose lower-priced plans or use computing-intensive features frequently. Business customers
• 13:34
  Speaker 1: may pay more for specialized applications, yet they will increasingly
• 13:38
  Speaker 1: expect clear evidence that these products improve productivity, reduce costs,
• 13:43
  Speaker 1: or generate additional income. The industry must ultimately demonstrate that
• 13:48
  Speaker 1: the value delivered to customers exceeds the costs of providing
• 13:52
  Speaker 1: the service. This challenge becomes particularly complicated as competition intensifies.
• 13:58
  Speaker 1: During the early stages of a new technology market, companies
• 14:02
  Speaker 1: may be able to charge premium prices because capable alternatives
• 14:07
  Speaker 1: are limited. As more developers enter the market and existing
• 14:10
  Speaker 1: competitors improve their products, customers gain greater bargaining power. Businesses
• 14:17
  Speaker 1: may then face pressure to reduce subscription prices, offer more
• 14:21
  Speaker 1: computing capacity for the same fee, or spend additional money
• 14:25
  Speaker 1: improving performance. Such competition can benefit consumers and accelerate adoption.
• 14:31
  Speaker 1: but it can also squeeze margins for companies that have
• 14:34
  Speaker 1: invested heavily in expensive infrastructure. Even when demand increases, profitability
• 14:39
  Speaker 1: may disappoint if prices fall faster than operating costs. In
• 14:44
  Speaker 1: other words, a rapidly growing AI market does not automatically
• 14:48
  Speaker 1: guarantee attractive returns for every business serving it. The spending
• 14:52
  Speaker 1: decisions of major technology companies add another layer of uncertainty.
• 14:57
  Speaker 1: Leading cloud providers and technology platforms are investing heavily in
• 15:02
  Speaker 1: computing capacity because they expect AI to become an essential
• 15:06
  Speaker 1: part of their products and services. Their investments may be
• 15:09
  Speaker 1: justified by a combination of direct AI revenue, increased customer retention,
• 15:15
  Speaker 1: and the strategic advantage of providing a broad technology ecosystem. However,
• 15:20
  Speaker 1: evaluating these commitments requires investors to distinguish between money spent
• 15:25
  Speaker 1: to generate new revenue and money spent primarily to defend
• 15:29
  Speaker 1: an existing market position. Companies may feel compelled to expand
• 15:34
  Speaker 1: even when immediate returns are difficult to establish, especially if
• 15:38
  Speaker 1: competitors are making similar moves. One of the most serious
• 15:42
  Speaker 1: risks emerging from the artificial intelligence investment boom is the
• 15:45
  Speaker 1: possibility that companies will build more computing capacity than the
• 15:50
  Speaker 1: market can profitably absorb. Across the technology sector, businesses are
• 15:55
  Speaker 1: investing heavily in graphics processing units, specialized AI chips, cloud infrastructure,
• 16:02
  Speaker 1: and enormous data centers in anticipation of years of expanding demand.
• 16:07
  Speaker 1: These investments are based on the expectation that artificial intelligence
• 16:12
  Speaker 1: will become increasingly important to businesses, governments, and consumers. However,
• 16:18
  Speaker 1: the construction of new infrastructure takes time, and companies often
• 16:22
  Speaker 1: make investment decisions using forecasts about demand that may change
• 16:26
  Speaker 1: before their projects are completed. If many businesses expand simultaneously
• 16:31
  Speaker 1: and their combined capacity grows faster than actual usage, the
• 16:36
  Speaker 1: industry could face an oversupply problem. Such a development would
• 16:40
  Speaker 1: not necessarily mean that artificial intelligence had failed, but it
• 16:44
  Speaker 1: could fundamentally change the economics of the market by placing
• 16:48
  Speaker 1: downward pressure on prices, utilization rates, and the profits that
• 16:53
  Speaker 1: investors expect from their investments. Citadel chief executive Ken Griffin
• 16:58
  Speaker 1: has drawn attention to this danger by comparing the economics
• 17:02
  Speaker 1: of AI computing capacity with with the economics of commercial airlines.
• 17:07
  Speaker 1: His argument centers on the difference between the enormous initial
• 17:10
  Speaker 1: cost of acquiring an asset and the relatively low additional
• 17:14
  Speaker 1: cost of using that asset once it is already available.
• 17:17
  Speaker 1: An airline invests heavily in aircraft, maintenance facilities, staff, and
• 17:23
  Speaker 1: other operating requirements before selling a single ticket. Once a
• 17:28
  Speaker 1: flight has been scheduled and most of those costs have
• 17:30
  Speaker 1: been committed, however, selling an otherwise empty seat at a
• 17:34
  Speaker 1: discount can still generate additional revenue. An airline may therefore
• 17:39
  Speaker 1: accept a low fare rather than allow a seat to
• 17:41
  Speaker 1: go unused, provided that the additional revenue exceeds the marginal
• 17:46
  Speaker 1: cost of carrying the passenger. Griffin sees a comparable dynamic
• 17:50
  Speaker 1: developing in AI computing, where companies that have already purchased
• 17:54
  Speaker 1: expensive processors may be willing to offer computing services at
• 17:58
  Speaker 1: increasingly low prices simply to keep their equipment operating and
• 18:02
  Speaker 1: generate some return from their investment. The comparison reveals why
• 18:07
  Speaker 1: an increase in demand does not automatically guarantee strong profits
• 18:11
  Speaker 1: for companies operating expensive infrastructure. Suppose a cloud provider spends
• 18:16
  Speaker 1: billions of dollars building a facility equipped with thousands of
• 18:19
  Speaker 1: AI processors. Management expects customers to purchase enough computing time
• 18:24
  Speaker 1: to cover operating costs, finance future upgrades, and eventually recover
• 18:30
  Speaker 1: the original investment. If demand develops as projected, the facility
• 18:35
  Speaker 1: may generate substantial revenue and attractive returns. But if several
• 18:40
  Speaker 1: competitors build similar facilities at the same time, customers gain
• 18:44
  Speaker 1: more choices, and some computing providers may find themselves competing
• 18:49
  Speaker 1: aggressively for business. Rather than leave processors idle, operators may
• 18:54
  Speaker 1: cut prices to attract customers and improve utilization. These discounts
• 18:59
  Speaker 1: can be commercially rational for individual companies because an idle
• 19:03
  Speaker 1: processor generates no revenue, while a discounted workload can still
• 19:07
  Speaker 1: contribute something toward electricity, staffing, maintenance, and other ongoing expenses. Yet,
• 19:15
  Speaker 1: when many providers make the same decision, the collective result
• 19:18
  Speaker 1: can be a price war that undermines the profitability of
• 19:21
  Speaker 1: the entire infrastructure market. The danger becomes greater when investors
• 19:26
  Speaker 1: confuse high demand for computing power with a guarantee that
• 19:30
  Speaker 1: computing providers will maintain their current pricing. AI adoption could
• 19:35
  Speaker 1: continue expanding rapidly while the cost of obtaining processing capacity
• 19:39
  Speaker 1: falls significantly. Businesses might use more computing resources than ever before,
• 19:45
  Speaker 1: yet suppliers could earn less revenue per unit of capacity
• 19:49
  Speaker 1: as competition intensifies this distinction matters because the financial success
• 19:54
  Speaker 1: of an infrastructure company depends not only on how much
• 19:58
  Speaker 1: equipment it owns or how many workloads it processes but
• 20:02
  Speaker 1: also on the prices it receives the efficiency with which
• 20:05
  Speaker 1: it operates and the amount of capital it must recover
• 20:09
  Speaker 1: A data center operating near full capacity can still disappoint
• 20:12
  Speaker 1: shareholders if its services are sold at prices too low
• 20:16
  Speaker 1: to generate an adequate return on the money invested in
• 20:19
  Speaker 1: the facility. Investors therefore face a difficult calculation. The expansion
• 20:24
  Speaker 1: of AI could be technologically successful and commercially widespread without
• 20:29
  Speaker 1: delivering the extraordinary profits that current market expectations may assume.
• 20:34
  Speaker 1: There is also a difference between the companies that manufacture
• 20:38
  Speaker 1: AI processors and the companies that purchase and operate them.
• 20:42
  Speaker 1: The debate surrounding Ken Griffin's warning about an artificial intelligence
• 20:46
  Speaker 1: bubble has revived one of the most important lessons in
• 20:50
  Speaker 1: modern financial history. A technology can transform the world while
• 20:54
  Speaker 1: delivering devastating losses to investors who pay too much for
• 20:58
  Speaker 1: its promise. To understand why this distinction matters, It is
• 21:02
  Speaker 1: necessary to look back at the dot-com boom of the
• 21:05
  Speaker 1: late 1990s, when the rapid expansion of the internet convinced
• 21:10
  Speaker 1: investors that a new economic era was beginning. Businesses were
• 21:14
  Speaker 1: discovering how digital communication could change commerce, advertising, banking, entertainment,
• 21:21
  Speaker 1: and the distribution of information. Entrepreneurs rushed to establish online companies,
• 21:27
  Speaker 1: venture capital flooded into technology start-ups, and stock prices climbed
• 21:32
  Speaker 1: as investors attempted to identify the businesses that would dominate
• 21:35
  Speaker 1: the emerging digital economy. Much of the excitement was justified
• 21:40
  Speaker 1: because the internet eventually became central to everyday life. Yet
• 21:44
  Speaker 1: the financial enthusiasm surrounding the technology became so intense that
• 21:49
  Speaker 1: market valuations increasingly reflected hopes about the future rather than
• 21:53
  Speaker 1: evidence of sustainable profits. The resulting collapse demonstrated that correctly
• 21:59
  Speaker 1: identifying a technological revolution does not necessarily mean correctly identifying
• 22:04
  Speaker 1: its financial winners. During the dot-com boom, investors frequently treated
• 22:10
  Speaker 1: an internet connection and a compelling growth story as evidence
• 22:14
  Speaker 1: that a company could become extraordinarily valuable. Traditional measures of
• 22:18
  Speaker 1: business performance, including profits, cash flow, and the ability to
• 22:23
  Speaker 1: cover operating expenses, sometimes received less attention than the potential
• 22:28
  Speaker 1: size of a company's future market. Businesses promised to attract
• 22:32
  Speaker 1: millions of users, reshape established industries, and establish dominant positions
• 22:38
  Speaker 1: in markets that were only beginning to develop. Some investors
• 22:42
  Speaker 1: believed that companies could eventually monetize their audiences after first
• 22:47
  Speaker 1: prioritizing growth, while others worried that those expectations were becoming
• 22:52
  Speaker 1: disconnected from commercial reality. In a rapidly changing industry, conventional
• 22:57
  Speaker 1: valuation methods appeared inadequate to some market participants, encouraging the
• 23:02
  Speaker 1: idea that businesses operating under the new digital model deserved
• 23:06
  Speaker 1: entirely different rules. Although certain internet companies developed lasting businesses,
• 23:12
  Speaker 1: many others struggled to turn attention and website traffic into
• 23:16
  Speaker 1: reliable revenue. When confidence weakened and investors reconsidered the financial
• 23:21
  Speaker 1: foundations of their investments, companies that had depended on continued
• 23:25
  Speaker 1: access to capital faced a severe test. The collapse of
• 23:29
  Speaker 1: technology stocks in the early 2000s exposed the consequences of
• 23:33
  Speaker 1: these assumptions. Investors who had purchased shares at exceptionally high
• 23:38
  Speaker 1: valuations discovered that even promising companies could lose substantial market
• 23:43
  Speaker 1: value when revenue growth disappointed or financing became harder to obtain.
• 23:49
  Speaker 1: Some businesses failed altogether, while others were acquired, reorganized, or
• 23:54
  Speaker 1: forced to reduce their operations. The impact extended beyond investors
• 23:59
  Speaker 1: holding individual technology stocks. Companies across the broader economy became
• 24:04
  Speaker 1: more cautious about spending, technology firms scaled back expansion plans,
• 24:09
  Speaker 1: and workers faced layoffs as businesses struggled to adapt to
• 24:13
  Speaker 1: the sudden change in financial conditions. The downturn also demonstrated
• 24:18
  Speaker 1: how investor confidence can influence decisions about hiring, infrastructure, research,
• 24:24
  Speaker 1: and corporate investment. When access to funding depends heavily on
• 24:28
  Speaker 1: optimistic market expectations, a sharp reversal in sentiment can force
• 24:33
  Speaker 1: businesses to change their plans regardless of how ambitious their
• 24:37
  Speaker 1: original objectives were. The internet itself did not disappear. but
• 24:41
  Speaker 1: the financial environment that had supported the most speculative parts
• 24:45
  Speaker 1: of the boom could no longer be sustained. The longer-term outcome, however,
• 24:50
  Speaker 1: was more complicated than a simple story of technological failure.
• 24:55
  Speaker 1: The internet continued developing after the bubble burst, and digital
• 24:59
  Speaker 1: services eventually transformed industries that had initially struggled to adapt.
• 25:04
  Speaker 1: Online shopping expanded, search engines became essential gateways to information,
• 25:10
  Speaker 1: digital advertising grew into a major business, and companies increasingly
• 25:14
  Speaker 1: moved important operations onto internet-based platforms. Businesses that survived and
• 25:20
  Speaker 1: developed durable commercial models gained opportunities to benefit from the infrastructure,
• 25:25
  Speaker 1: consumer habits, and technological capabilities established during the earlier boom. Meanwhile,
• 25:32
  Speaker 1: Assets and expertise created during the expansion could continue supporting
• 25:37
  Speaker 1: economic activity even after some of the companies responsible for
• 25:40
  Speaker 1: financing them had disappeared. This is one reason the dot-com
• 25:44
  Speaker 1: experience remains relevant to the AI debate. The failure of
• 25:48
  Speaker 1: investors or businesses during a technological transition does not necessarily
• 25:53
  Speaker 1: invalidate the technology's long-term economic value. The growing debate over
• 25:59
  Speaker 1: an artificial intelligence bubble is beginning to have consequences beyond
• 26:03
  Speaker 1: financial forecasts and corporate investment announcements, as investors increasingly question
• 26:09
  Speaker 1: whether the enormous sums being committed to the technology will
• 26:13
  Speaker 1: produce the returns they expect. For much of the recent
• 26:16
  Speaker 1: AI boom, optimism about future growth helped propel technology companies
• 26:22
  Speaker 1: and semiconductor manufacturers to extraordinary market valuations. Investors were willing
• 26:28
  Speaker 1: to pay premium prices for businesses supplying the chips, computing infrastructure, software,
• 26:34
  Speaker 1: and cloud services needed to develop advanced AI systems. However,
• 26:38
  Speaker 1: by October 2026, that confidence was being tested by renewed
• 26:43
  Speaker 1: volatility across technology markets, growing concerns about financing costs and
• 26:48
  Speaker 1: uncertainty surrounding the commercial returns on AI infrastructure. The resulting
• 26:53
  Speaker 1: turbulence has brought a more difficult question to the forefront
• 26:57
  Speaker 1: of Wall Street. Can companies continue expanding their AI operations
• 27:01
  Speaker 1: at the current pace, while satisfying investors who increasingly want
• 27:05
  Speaker 1: proof that spending is translating into sustainable profits? The reassessment
• 27:10
  Speaker 1: has been particularly visible in semiconductor shares, which have become
• 27:15
  Speaker 1: closely associated with expectations for continued AI growth. Companies producing
• 27:20
  Speaker 1: advanced processors and memory components have benefited from strong demand
• 27:25
  Speaker 1: as technology businesses compete to develop more capable systems and
• 27:30
  Speaker 1: expand their computing capacity. Their products are essential to the
• 27:34
  Speaker 1: operation of modern AI models, giving manufacturers an important position
• 27:39
  Speaker 1: within the industry's supply chain. Yet that strategic importance has
• 27:44
  Speaker 1: not insulated their stocks from market volatility. During the middle
• 27:48
  Speaker 1: of October, semiconductor shares experienced significant selling pressure, contributing to
• 27:54
  Speaker 1: a wider retreat in technology markets. Some investors began questioning
• 27:58
  Speaker 1: whether valuations had moved too far ahead of the industry's
• 28:02
  Speaker 1: long-term earnings potential, while others worried that the competition to
• 28:06
  Speaker 1: supply AI infrastructure could eventually place pressure on prices and profitability.
• 28:12
  Speaker 1: These concerns do not necessarily indicate that demand for chips
• 28:16
  Speaker 1: is disappearing. Instead, They reflect the difficulty of determining how
• 28:20
  Speaker 1: much future growth is already incorporated into current share prices.
• 28:25
  Speaker 1: The pressure also spread beyond chip makers to businesses providing
• 28:29
  Speaker 1: specialized cloud computing and AI infrastructure. These companies occupy an
• 28:34
  Speaker 1: important position in the technology ecosystem because they rent computing
• 28:39
  Speaker 1: power to businesses that need access to advanced processors without
• 28:43
  Speaker 1: necessarily building their own facilities. The business model can generate
• 28:48
  Speaker 1: substantial revenue when customer demand is strong and infrastructure is
• 28:52
  Speaker 1: used efficiently. However, it can also involve considerable financial risk
• 28:57
  Speaker 1: when expansion requires large amounts of borrowed money. Companies building
• 29:02
  Speaker 1: new data centers must pay for equipment, land, electricity infrastructure, construction,
• 29:08
  Speaker 1: cooling systems, and networking before they can collect all the
• 29:12
  Speaker 1: revenue expected from future customers. When financing conditions become less
• 29:17
  Speaker 1: favorable or investor confidence weakens, those commitments receive greater scrutiny.
• 29:23
  Speaker 1: Recent market pressure on AI infrastructure providers has therefore highlighted
• 29:27
  Speaker 1: a growing distinction between companies that generate substantial cash from
• 29:32
  Speaker 1: established operations and those whose expansion depends heavily on continued
• 29:37
  Speaker 1: borrowing and optimistic projections of future. demand interest rates have
• 29:42
  Speaker 1: become another important factor influencing how investors evaluate the ai
• 29:46
  Speaker 1: sector technology companies often receive high valuations because markets expect
• 29:52
  Speaker 1: them to generate substantial profits in the future the further
• 29:55
  Speaker 1: those expected profits lie in the future the more sensitive
• 29:59
  Speaker 1: their estimated present value can become to changes in interest
• 30:03
  Speaker 1: rates and financing assumptions. When inflation remains stubborn and investors
• 30:08
  Speaker 1: begin anticipating that central banks may keep borrowing costs elevated,
• 30:13
  Speaker 1: the financial calculations supporting expensive growth investments can change quickly.
• 30:18
  Speaker 1: Businesses financing data centers with debt may have to pay
• 30:22
  Speaker 1: more to raise capital, while investors may become less willing
• 30:26
  Speaker 1: to accept high valuations for companies whose profitability remains uncertain.
• 30:31
  Speaker 1: Higher financing costs can also affect customers, potentially slowing their
• 30:35
  Speaker 1: willingness to commit to expensive technology projects. This creates a
• 30:40
  Speaker 1: difficult environment for businesses that need to keep investing aggressively
• 30:44
  Speaker 1: to preserve their competitive positions while facing increasing pressure to
• 30:48
  Speaker 1: demonstrate financial discipline. The connection between inflation and AI investment
• 30:54
  Speaker 1: is particularly important because it affects more than the immediate
• 30:58
  Speaker 1: cost of borrowing. Persistent inflation can increase construction expenses, equipment prices,
• 31:04
  Speaker 1: labor costs, and the cost of securing reliable electricity supplies.
• 31:09
  Speaker 1: The debate over an artificial intelligence bubble is no longer
• 31:13
  Speaker 1: limited to stock prices, semiconductor companies, or the enormous sums
• 31:18
  Speaker 1: being spent on data centers. If the investment boom experiences
• 31:22
  Speaker 1: a significant correction, the consequences could spread into employment, business expansion, entrepreneurship,
• 31:29
  Speaker 1: public finances, and the everyday lives of workers whose jobs
• 31:33
  Speaker 1: are being reshaped by automation. Artificial intelligence is already influencing
• 31:39
  Speaker 1: decisions about how companies organize their operations, allocate budgets, recruit employees,
• 31:45
  Speaker 1: and plan for future growth. A sharp reversal in investor
• 31:49
  Speaker 1: confidence could change the pace of that transformation, particularly if
• 31:53
  Speaker 1: companies become more cautious about spending and lenders tighten access
• 31:57
  Speaker 1: to financing. At the same time, a correction could make
• 32:01
  Speaker 1: computing resources less expensive and create opportunities for businesses that
• 32:06
  Speaker 1: were previously unable to afford advanced AI systems. The potential
• 32:11
  Speaker 1: outcome is therefore complicated. A financial setback for some companies
• 32:15
  Speaker 1: could coexist with broader access to useful technology. while workers
• 32:20
  Speaker 1: and communities experience both new opportunities and significant uncertainty. One
• 32:26
  Speaker 1: of the most immediate risks involves employment in the technology
• 32:30
  Speaker 1: industry itself. During periods of strong investment, companies hire engineers, researchers,
• 32:37
  Speaker 1: software developers, construction workers, data center technicians, and specialists responsible
• 32:43
  Speaker 1: for maintaining the infrastructure supporting AI systems. The construction of
• 32:47
  Speaker 1: large computing facilities also creates demand for electricians, equipment suppliers,
• 32:53
  Speaker 1: cooling specialists, transportation services, and other businesses connected to infrastructure development.
• 33:00
  Speaker 1: If financing becomes more expensive or demand for computing capacity
• 33:04
  Speaker 1: falls below expectations, some projects could be delayed, reduced, or canceled.
• 33:11
  Speaker 1: Companies that had recruited staff in anticipation of continued expansion
• 33:15
  Speaker 1: might then reconsider their hiring plans, freeze vacancies, or reduce
• 33:20
  Speaker 1: their workforces. The impact would extend beyond highly paid technology
• 33:24
  Speaker 1: professionals because major infrastructure projects support employment across a broad
• 33:29
  Speaker 1: network of contractors and suppliers. Communities that have invested in
• 33:34
  Speaker 1: attracting data centers and related businesses could also face disappointing
• 33:39
  Speaker 1: economic returns if anticipated projects fail to proceed as planned.
• 33:44
  Speaker 1: The consequences would not necessarily be limited to layoffs. A
• 33:48
  Speaker 1: weaker investment environment could change the kinds of jobs companies
• 33:52
  Speaker 1: are willing to create and the skills they prioritize when recruiting.
• 33:56
  Speaker 1: Businesses might concentrate on applications with clear commercial benefits rather
• 34:01
  Speaker 1: than experimental projects whose financial returns remain uncertain. Research teams
• 34:07
  Speaker 1: could face pressure to demonstrate practical results, while startups might
• 34:11
  Speaker 1: shift from developing expensive foundational models toward building applications on
• 34:16
  Speaker 1: existing platforms. Some companies could continue hiring specialists who help
• 34:21
  Speaker 1: integrate AI into established operations and even as others reduce
• 34:26
  Speaker 1: recruitment in areas where spending has slowed. This would create
• 34:30
  Speaker 1: an uneven employment landscape in which demand for certain technical
• 34:34
  Speaker 1: skills remains strong while opportunities in more speculative parts of
• 34:38
  Speaker 1: the industry become less reliable. Workers entering technology careers would
• 34:43
  Speaker 1: need to consider not only whether AI is expanding, but
• 34:46
  Speaker 1: also which parts of the industry can maintain sustainable businesses
• 34:50
  Speaker 1: when investor enthusiasm becomes more restrained. A correction could also
• 34:55
  Speaker 1: accelerate changes in the relationship between artificial intelligence and the
• 34:59
  Speaker 1: wider workforce. Companies have been exploring AI tools as a
• 35:04
  Speaker 1: way to automate repetitive tasks, improve productivity, and reduce the
• 35:08
  Speaker 1: cost of certain business operations. During a period of financial pressure,
• 35:13
  Speaker 1: those objectives could become even more important as executives search
• 35:17
  Speaker 1: for ways to protect profit margins. Some employers might increase
• 35:22
  Speaker 1: their use of AI to handle administrative work, generate reports,
• 35:26
  Speaker 1: support customer service, or assist employees with routine analysis. However,
• 35:32
  Speaker 1: financial uncertainty could also lead other organizations to postpone automation
• 35:36
  Speaker 1: projects because implementation requires spending on software, staff training, security,
• 35:43
  Speaker 1: and integration with existing systems. the result would not be
• 35:47
  Speaker 1: a uniform acceleration or reversal of automation. Instead, businesses would
• 35:52
  Speaker 1: make decisions based on their financial resources, the practical usefulness
• 35:57
  Speaker 1: of AI tools, and the urgency of reducing operating costs.
• 36:02
  Speaker 1: For workers, the central concern is that technological change could
• 36:06
  Speaker 1: continue even if the companies financing it experience a financial correction.
• 36:10
  Speaker 1: A decline in AI valuations would not erase the software
• 36:14
  Speaker 1: systems already developed or eliminate the potential productivity benefits they offer.
• 36:19
  Speaker 1: Businesses that have successfully integrated AI into their operations may
• 36:23
  Speaker 1: continue using it because the tools help them complete tasks
• 36:27
  Speaker 1: more efficiently. The future of artificial intelligence investment now depends
• 36:31
  Speaker 1: on a question that reaches far beyond the rise and
• 36:34
  Speaker 1: fall of individual stock prices. Can the technology deliver enough
• 36:39
  Speaker 1: lasting economic value to justify the extraordinary financial commitments being
• 36:44
  Speaker 1: made in its name? Ken Griffin's warning about an AI
• 36:47
  Speaker 1: bubble has brought renewed attention to the gap between what
• 36:50
  Speaker 1: investors expect and what businesses can realistically earn. Artificial intelligence
• 36:56
  Speaker 1: has the potential to transform industries, improve productivity, accelerate scientific research,
• 37:02
  Speaker 1: and create entirely new markets. Yet these possibilities do not
• 37:06
  Speaker 1: guarantee that every company building AI systems, supplying computing infrastructure,
• 37:12
  Speaker 1: or financing new developments will generate an adequate return. As
• 37:16
  Speaker 1: the industry moves into its next phase, investors will need
• 37:20
  Speaker 1: to distinguish between companies with sustainable business models and those
• 37:25
  Speaker 1: whose valuations depend heavily on continued enthusiasm. The outcome will
• 37:30
  Speaker 1: influence not only technology markets, but also corporate spending, employment, competition,
• 37:36
  Speaker 1: and the pace at which AI becomes part of everyday
• 37:39
  Speaker 1: economic activity. One possible outcome is that the industry gradually
• 37:44
  Speaker 1: grows into its current investment expectations. Under this scenario, businesses
• 37:49
  Speaker 1: discover increasingly valuable ways to incorporate AI into their operations,
• 37:54
  Speaker 1: allowing the revenue generated by AI services to catch up
• 37:58
  Speaker 1: with the enormous costs of developing and deploying them. Companies
• 38:02
  Speaker 1: could use AI to reduce administrative expenses, accelerate product development,
• 38:07
  Speaker 1: improve logistics, strengthen customer support, and help employees complete complex
• 38:13
  Speaker 1: tasks more efficiently. If these improvements translate into measurable financial benefits,
• 38:19
  Speaker 1: businesses may become more willing to pay for advanced AI
• 38:22
  Speaker 1: tools and commit to long-term contracts with providers. Cloud computing
• 38:27
  Speaker 1: companies could then generate more revenue from their infrastructure and
• 38:31
  Speaker 1: while software developers could build profitable services around increasingly capable models.
• 38:37
  Speaker 1: This outcome would support continued investment, although returns would probably
• 38:42
  Speaker 1: differ substantially between companies depending on their operating costs, competitive positions,
• 38:47
  Speaker 1: and ability to retain customers. A second possibility is that
• 38:52
  Speaker 1: AI adoption continues growing, but the financial returns fall short
• 38:56
  Speaker 1: of today's most ambitious expectations. Companies may find useful applications
• 39:02
  Speaker 1: across many industries while discovering that customers are unwilling to
• 39:06
  Speaker 1: pay premium prices for every new capability. As competing developers
• 39:11
  Speaker 1: introduce alternative models and services, businesses may gain greater bargaining
• 39:16
  Speaker 1: power and push prices downward. The cost of computing could decline,
• 39:21
  Speaker 1: making AI more accessible while simultaneously reducing the profit margins
• 39:25
  Speaker 1: of infrastructure providers. Some businesses might use AI primarily to
• 39:30
  Speaker 1: improve existing products rather than purchase entirely new services limiting
• 39:35
  Speaker 1: the additional revenue available to technology suppliers under these circumstances
• 39:41
  Speaker 1: artificial intelligence could still become an essential part of the
• 39:44
  Speaker 1: economy Yet investors who purchase shares at exceptionally high valuations
• 39:49
  Speaker 1: might experience substantial losses. The technology would continue progressing, but
• 39:55
  Speaker 1: the financial market would need to adjust its expectations to
• 39:58
  Speaker 1: reflect more modest margins and a slower recovery of invested capital.
• 40:03
  Speaker 1: A third scenario involves a more abrupt correction triggered by
• 40:07
  Speaker 1: excess computing capacity, rising financing costs, or disappointing commercial results.
• 40:13
  Speaker 1: If companies expand infrastructure faster than customers can absorb it,
• 40:18
  Speaker 1: providers may compete aggressively by lowering prices to keep their
• 40:22
  Speaker 1: equipment operating. Revenue per unit of computing capacity could fall,
• 40:26
  Speaker 1: while expenses associated with construction, electricity, maintenance, and debt remain significant.
• 40:33
  Speaker 1: Businesses that relied on high utilization rates and optimistic pricing
• 40:38
  Speaker 1: assumptions could then struggle to generate sufficient cash to cover
• 40:42
  Speaker 1: their commitments. Investors might respond by reducing valuations, lenders could
• 40:48
  Speaker 1: tighten financing conditions, and some planned data center projects could
• 40:52
  Speaker 1: be postponed or canceled. Startups dependent on repeated fundraising rounds
• 40:58
  Speaker 1: would face particular pressure, while established companies might abandon projects
• 41:03
  Speaker 1: that no longer meet their financial requirements. Such a correction
• 41:07
  Speaker 1: could be painful, but it would not necessarily mark the
• 41:10
  Speaker 1: end of the AI revolution. It could instead represent a
• 41:13
  Speaker 1: period in which investment levels, infrastructure capacity, and market valuations
• 41:19
  Speaker 1: adjust to more realistic estimates of demand. Determining which scenario
• 41:23
  Speaker 1: is developing will require investors to look beyond ambitious announcements
• 41:27
  Speaker 1: and headline spending figures. One important indicator will be the
• 41:31
  Speaker 1: relationship between AI revenue and the capital expenditure required to
• 41:36
  Speaker 1: produce it.