The same week brought two seemingly contradictory pieces of news. Nvidia halved its financing guarantee for OpenAI’s massive planned data center in Ohio after investors pushed back on the chipmaker’s risk exposure. At the same time, Anthropic reported a revenue jump that nearly tripled its prior quarter and is preparing an IPO at a valuation approaching a trillion dollars. Both stories feed the same debate: is the AI industry in a bubble, or is this growth real? A closer look suggests both answers hold some truth.
Key takeaways
- Nvidia halved its financing guarantee for OpenAI’s planned Ohio data center from $250 billion to roughly $120 billion, according to the Wall Street Journal, after its own investors raised concerns about risk concentration.
- The reduced guarantee covers only the first construction phase of 5 of the planned 10 gigawatts, OpenAI is negotiating the rest separately with SoftBank subsidiary SB Energy.
- Anthropic reported revenue jumping from $4.73 billion in Q1 to more than $11.5 billion in Q2 2026.
- Anthropic is preparing an IPO for late September or early October, with investors discussing valuations between roughly $965 billion and $2 trillion.
- Critics note that at a $2 trillion valuation, Anthropic would need annual net income of $59 to 79 billion to meet typical Nasdaq-100 benchmarks, and the company is currently barely profitable.
Why Nvidia is trimming its own risk
Nvidia originally planned to guarantee up to $250 billion for OpenAI’s Ohio megaproject, a key piece of the 10-gigawatt data center campus being built by SoftBank subsidiary SB Energy. According to the Wall Street Journal, Nvidia has now cut that commitment to just under $120 billion, covering only the first roughly 5-gigawatt construction phase. OpenAI is negotiating a separate lease for the rest on its own. Nvidia’s stock dropped about 5 percent on the news. The move can be read two ways: as a cautious signal that even the biggest beneficiary of the AI boom no longer wants to concentrate balance-sheet risk on a single project, or simply as a routine restructuring of a mega-deal now backed by a broader consortium of financial investors including Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR. The two readings aren’t mutually exclusive, but the fact that it was Nvidia’s own investors pushing for caution is exactly the kind of data point AI bubble skeptics like to cite.
Anthropic’s counter-evidence: growth that looks like real demand
While Nvidia turns more cautious, Anthropic is posting numbers that look more like genuine market demand than pure hype. Revenue rose from $4.73 billion in Q1 to more than $11.5 billion in Q2 2026, a roughly 140 percent jump in three months. Anthropic itself projects annual revenue of $190 to $200 billion by 2028. On the strength of that trajectory, the company is preparing an IPO for late September or early October. While a valuation circulating in May 2026 stood at around $965 billion, some investors are now modeling figures as high as $2 trillion. That growth rate is remarkable even by the AI industry’s already frantic standards, and it’s hard to explain through speculation alone, it reflects paying customer demand for Claude models across enterprise software, coding tools, and, as a recent article on Anthropic’s security research shows, increasingly demanding technical research fields as well.
The math that has to justify the valuation
This is exactly where the criticism lands. At a $2 trillion valuation, Anthropic would need annual net income between $59 and $79 billion under typical Nasdaq-100 standards. The company is far from that today, in Q2 2026 it was expected to post its first operating profit, but operating profit isn’t the same as net income, taxes and debt interest can still weigh heavily on the bottom line. Renaissance Capital analyst Avery Marquez frames the approaching profitability as at least something that could make the massive valuation “maybe not seem so crazy.” Tellingly, the frequent comparison to Amazon, which earns roughly $77.7 billion in net income, cuts both ways, a significant chunk of Amazon’s recent quarterly profit itself came from its own investment stake in Anthropic. It’s a circular argument that doesn’t make the debate any simpler, and it shows just how tightly the valuations of major cloud companies and their AI stakes are now intertwined. Investor Evan Schlossman of Neostellar Capital instead points to globally scarce compute capacity as the real value driver, as long as demand for compute outstrips supply, he argues, Anthropic and OpenAI can both thrive without their valuations canceling each other out.
Bubble or not: a question of timing
The two stories this week contradict each other less than they first appear to. Nvidia is diversifying its risk without exiting the AI buildout, the company remains deeply embedded in OpenAI’s infrastructure through a multibillion-dollar commitment. Anthropic’s revenue growth is real and unusually strong, but it doesn’t yet prove that the trillion-dollar valuations under discussion are sustainable. Anyone waiting for one clear signal for or against an AI bubble will likely be waiting a while longer. A more realistic picture is a market correcting itself even as it keeps growing, capital providers are spreading risk more broadly, analysts are scrutinizing valuations more strictly, and individual companies like Anthropic now actually have to turn growth into profit. For observers, the more useful question over the coming months isn’t “bubble, yes or no,” but whether Anthropic’s revenue explosion actually converts into sustainable profitability, or remains another impressive but unprofitable chapter of the AI gold rush.