Op-ed urges FTC to unwind Microsoft-OpenAI crossholdings

The pitch from Sumit Sharma at Tech Policy Press is that the FTC does not need to wait for a market crash to matter, because the plumbing of the AI bubble is antitrust-shaped already. The 'Magnificent Seven' tech stocks are now roughly 35 percent of the S&P 500, and AI-related capital expenditure accounted for more than 90 percent of US GDP growth in the first half of 2025. Under all of that sit two circular money loops linking the hyperscalers to the frontier labs.

The first is what Sharma calls a cloud-for-credit exchange, where Microsoft, Google, and Amazon invest in AI companies like OpenAI and Anthropic and then book the labs' returning cloud spend as their own revenue. The second is mark-to-market accounting on the equity itself. Amazon's stake in Anthropic alone added $16.8 billion to its Q1 2026 earnings, per the piece, and Alphabet reported roughly $28.7 billion in similar unrealized gains. Neither number reflects a customer paying a bill; both flow to the bottom line.

Microsoft-OpenAI is the case study. Microsoft owns roughly 27 percent of OpenAI Group PBC, a stake valued at $135 billion, and can use OpenAI's IP through 2032 and share in revenue payments through 2030. Sharma cites internal Microsoft correspondence framing the 'worst case scenario' as OpenAI ditching Azure for AWS, which reads more like competitor management than a passive minority position. The FTC's January 2025 6(b) Staff Report on AI Partnerships and Investments already laid out how these cloud-for-equity contracts bind labs to their would-be competitors.

The spillover into the real economy runs through utilities and retail investors. Sharma points to a proposed $420 billion mega-merger between Dominion Energy and NextEra premised on speculative data-center demand, with Virginia power costs up 12 percent since February 2025, and to an Allbirds rebrand as an AI company after which the stock surged 350 percent in a single week. His remedy list is ring-fencing regulated utilities from AI data-center bets, unwinding crossholdings and board interlocks, and applying the 2023 US Merger Guidelines to minority stakes.

The honest caveat is that this is an op-ed making a policy case, not reporting on filings that have actually been brought; the article maps mechanisms and precedents rather than naming which specific enforcement actions the FTC has already opened. What it does not give you is a divestiture path, how you unwind a 27 percent stake and years of IP entanglement without triggering the market correction it warns against. For independent labs, smaller cloud providers, and state utility regulators, though, the strategic read is that the antitrust toolkit is being sharpened while the equity math still flatters the incumbents.