Michigan economist Patrick Anderson warns that retirement investors could face losses if the AI investment boom suffers a correction similar to the dot-com crash of 2000.
LANSING — Michigan’s public retirement system held approximately $5.3 billion in six major technology companies involved in artificial intelligence at the end of September, raising concerns about the potential impact on retirement investments if the AI stock market boom suffers a major correction.
According to the State of Michigan Retirement System’s September 30 securities disclosure, filed with the SEC on October 27, the holdings included approximately $1.6 billion in Nvidia, $1.05 billion in Alphabet, $987 million in Microsoft, $643 million in Amazon, $568 million in Broadcom and $451 million in Advanced Micro Devices.
Together, those investments represented approximately 23% of the retirement system’s $22.6 billion in reported Form 13F securities. The filing covers reportable securities, not the retirement system’s entire investment portfolio.
These companies have substantial businesses beyond AI, and the figures do not establish direct investments in Anthropic.
Nevertheless, the holdings demonstrate how closely Michigan retirement investments are tied to technology companies benefiting from the AI boom.
What This Means for Your Retirement Savings
- $5.3 billion: Michigan public retirement holdings in six major AI-related technology companies as of September 30.
- 23%: Their combined share of the system’s reported Form 13F securities—not its total retirement assets.
- 401(k)s and IRAs: Investors holding broad stock index funds may already own many of these companies.
- 77%: The Nasdaq’s approximate decline during the 2000–2002 dot-com crash.
- The warning: AI could continue transforming the economy even if investors suffer substantial stock market losses.
Michigan Economist Warns Retirement Investors
The potential risks are drawing fresh attention following Anthropic’s October 26 IPO registration.
Anthropic, developer of the Claude AI platform, reported $16.3 billion in revenue during the first six months of 2026 but an $80.7 billion net loss, largely reflecting noncash accounting charges.
Patrick Anderson, founder of Michigan-based Anderson Economic Group, says ordinary retirement savers already face exposure to the AI investment boom, whether or not they intend to purchase Anthropic shares.
“Almost every worker that has spent a decade or more building up retirement savings already has exposure to ‘AI’ stocks. Tech companies have become such a large share of the overall stock market capitalization that even an investor that owns just index securities is already exposed to the gyrations of the tech economy,” Anderson said.
“For this reason, I think even investors that don’t intend to buy Anthropic or OpenAI stock directly should spend some time looking at the Anthropic investor disclosure.”
Anthropic’s $80 Billion Loss Doesn’t Tell the Whole Story
According to Anthropic’s SEC registration statement, first-half revenue climbed from approximately $1.2 billion in 2025 to $16.3 billion in 2026.
Its operating loss narrowed to approximately $693 million, compared with $2.1 billion a year earlier.
Much of the reported $80.7 billion net loss involved accounting adjustments associated with convertible financing instruments, including approximately $72.4 billion in noncash charges.
That means Anthropic did not lose $80 billion through ordinary business operations.
However, the company faces enormous future financial commitments.
Its SEC filing identifies approximately $321 billion in future payments for leases not yet commenced and another $266 billion in minimum purchase commitments as of June 30.
Additional agreements and amendments have subsequently been disclosed. Because some commitments may overlap, the figures should not simply be added together.
The central investment question is whether Anthropic can generate enough future revenue and cash flow to support its expansion.
Is AI Following the Dot-Com Bubble of 2000?
The enormous investments and valuations surrounding AI companies are reviving memories of the internet boom that ended in a devastating stock market collapse.
During the late 1990s, investors correctly anticipated that the internet would transform business and communications.
However, many paid extraordinary prices for companies that never generated sustainable profits.
The Nasdaq Composite subsequently plunged approximately 77% between March 2000 and October 2002, as documented in National Bureau of Economic Research analysis of the technology-stock collapse.
Numerous internet companies failed, and investors lost substantial wealth.
Yet the internet revolution continued, eventually transforming retail, banking, manufacturing and communications.
Anderson believes a similar distinction exists between AI technology and the sustainability of today’s investment boom.
“AI, especially the large language models that underly Claude, ChatGPT, and others, are here to stay. There are serious questions about the sustainability of the growth plans already announced by chip makers, data centers, and AI vendors. However, I don’t see how AI as a whole stops being part of the software and communications technology we use every day.”
Unlike many speculative internet startups in 2000, today’s AI expansion also involves established, profitable technology companies.
But even financially strong businesses can overinvest if anticipated demand fails to materialize.
Michigan’s Financial Exposure Extends Beyond Stocks
Anderson warns that a significant AI investment downturn could affect companies involved in building and powering data centers.
“There are real risks with AI investments, and they extend beyond providers like Anthropic and OpenAI to chipmakers like Nvidia, data center operators, and even some real estate and power producers. Michigan is just as vulnerable to these as other states.”
Michigan is attracting major data center proposals that require substantial investments in electricity infrastructure.
If demand for AI computing grows more slowly than anticipated, developers could delay construction, reduce expansion plans or cancel projects.
However, Michigan regulators have adopted safeguards intended to protect existing electricity customers from some of those financial risks.
In November 2025, the Michigan Public Service Commission approved special requirements for large Consumers Energy customers.
The provisions include minimum 15-year contracts, billing based on at least 80% of contracted electricity demand, early termination fees and financial security requirements.
The Commission also approved special electricity arrangements involving DTE Electric and Google’s proposed Van Buren Township data center in October 2026, including long-term contractual and financial protections.
Those safeguards are designed to reduce the risk that households and businesses will bear infrastructure costs if large customers fail to meet their commitments.
Whether they would fully protect ratepayers during a severe industry downturn—including a major customer’s bankruptcy or default—remains an important question.
MITechNews contacted the Michigan Public Service Commission seeking clarification on whether existing financial safeguards would fully protect electricity customers if a major data center operator declared bankruptcy, defaulted on its contract or substantially reduced electricity consumption. The Commission had not responded by publication time.
Could AI Survive While Investors Lose Money?
Anthropic’s IPO could become a major test of Wall Street’s willingness to finance the extraordinary expansion of artificial intelligence.
Its financial disclosures demonstrate substantial commercial demand, but also illustrate the enormous costs and commitments involved.
For Michigan, the consequences extend beyond one company’s stock price.
A sharp decline in AI investment could affect retirement portfolios, technology employment, construction projects and economic development.
At the same time, cheaper AI technology could benefit Michigan manufacturers, healthcare providers and small businesses seeking productivity improvements.
The dot-com crash demonstrated that a technological revolution can succeed even when investors suffer devastating financial losses.
Anderson believes artificial intelligence is here to stay.
The unanswered question is whether today’s enormous AI stock valuations and infrastructure commitments can ultimately be justified by sustainable profits.
