OpenAI's darkest hour: Revenue feared to be cut by 70%, how much longer can a valuation of hundreds of billions hold?

Author: Scott Galloway & Ed Elson

Compiled by: Shenchao TechFlow

Senchao Guide: Apple sues, Oracle downgrades ratings, price wars begin------OpenAI has just experienced the worst week in its serious history. Even worse, if all these risks materialize, its revenue forecast for 2030 could plummet by 70%, with cash flow losses reaching $165 billion. Will this $100 billion-valued AI giant become the largest tech bubble in history?

OpenAI may not achieve its 2030 revenue forecast of 70%, here's why

It has been another terrible week for OpenAI. The company was reported to have sold advanced AI models to Chinese companies on the Pentagon's blacklist, the first AI device was leaked (allegedly a portable speaker), and according to Emarketer's latest forecast, OpenAI's advertising business is expected to be 95% lower than its own predictions.

And that's not all. Apple sued OpenAI last week, accusing it of stealing intellectual property in its consumer hardware plans. S&P Global Ratings also downgraded Oracle's debt to BBB-, just one notch above junk status, citing that OpenAI constitutes a "key credit risk." Additionally, DeepSeek is reportedly preparing for an IPO, with an application possibly submitted as early as this year. A cheaper Chinese AI model provider successfully went public, which may make it harder for OpenAI and Anthropic to attract funding.

Overall, these issues raise questions about whether OpenAI can meet its revenue forecasts and fulfill its obligations under contracts worth hundreds of billions with computing power suppliers and chip companies.

First, Apple's lawsuit could bring OpenAI's entire hardware business to a standstill. Apple accuses OpenAI of poaching over 400 Apple employees, extracting confidential information from them, and then enticing Apple's suppliers to do proprietary work for OpenAI without permission. Apple is seeking monetary damages and an order for OpenAI to return or destroy all misappropriated property.

Second, the AI price war has begun, with companies like DeepSeek posing the biggest threat. Open-source Chinese models now account for nearly 50% of enterprise token usage on OpenRouter (an AI model marketplace). This was only 4.5% in the first half of 2025.

In response, American companies are significantly lowering prices. Last week, Meta announced the launch of a new model, Muse Spark 1.1, priced 75% lower than OpenAI and Anthropic. Under industry pressure, OpenAI released a model priced 80% lower than its own.

In the worst-case scenario, if Apple's lawsuit shuts down OpenAI's hardware business, ChatGPT's advertising revenue is as dismal as EMarketer predicts, and the price war forces OpenAI to cut model prices by 80%, then OpenAI's revenue will decline by 40% in 2026 and by 70% in 2030.

For a company that, in an ideal scenario, can only cover about 80% of its cash consumption by 2030, this situation would be catastrophic.

This would also affect when OpenAI achieves positive cash flow. According to internal forecasts, OpenAI is expected to achieve positive cash flow in 2030. However, in this downturn scenario, it would instead incur a loss of $165 billion that year.

OpenAI CEO Sam Altman attempted to reassure investors with a tweet, but his statement ultimately only promised to "do the right thing." Whatever that means.

The best business model in history is stealing intellectual property. The second best is: providing 80% of the value of a product at half the price. This is exactly what DeepSeek and other Chinese open-source weight models are trying to do now.

The U.S. has placed a huge bet on AI, while China has just rolled out a product that is close to the cutting edge at a fraction of the cost. Once Trump figures out what’s happening, this will become the next geopolitical football.

The market has not broadened------it has just become better at hiding AI

Investors have been hearing that the stock market is broadening. But is that really the case? The deeper you look, the harder it is to argue that stocks, bonds, or even alternative assets are now a big bet on AI.

This pattern is most evident in the stock market. AI-related stocks account for over 50% of the S&P 500 index by weight, and if AI and energy were removed from the S&P 500 this year, the S&P 500 would be negative.

AI is the hidden catalyst driving returns in seemingly unrelated sectors. For example, three of the four best-performing companies in the S&P 500 real estate sector are real estate investment trusts (REITs) focused on developing AI data centers.

Utility companies are benefiting from the soaring electricity demand driven by AI. Last year, U.S. electricity demand surged to an all-time high, with data centers accounting for about 50% of the demand growth.

Industrial stocks have soared due to the construction demand for AI data centers. In fact, for the first time since 2021, the forward P/E ratio of S&P 500 industrial stocks (26 times) is higher than that of tech companies (24 times).

The financial sector also relies on AI. Major banks are collecting record fees from AI company IPOs and M&A activity, as well as generating record trading revenue from market speculation around AI. Financial Times' Robert Armstrong even wrote, "It’s not an exaggeration to summarize: major banks are now direct investment targets in AI."

Even the Russell 2000 small-cap index saw 52% of its returns in the first half of this year come from AI-related companies.

Emerging markets are no exception. South Korea and Taiwan account for 75% of emerging market returns, with most of these gains coming from three AI semiconductor suppliers: TSMC, Samsung, and SK Hynix.

In Europe, just nine AI winners account for about 47% of this year's returns in the Stoxx Europe 600 index.

Apollo's chief economist Torsten Slok succinctly articulated the implications of this dependence: "This AI thing better succeed."

Real Estate Investment Trusts (REITs) are companies that own, operate, or finance real estate—apartments, hotels, or increasingly, data centers. Many REITs are publicly traded like stocks, so buying a share means buying into a professionally managed real estate portfolio. REITs are required to distribute at least 90% of their annual taxable income as dividends to shareholders.

CNBC experts hold stocks, so they always find reasons for others to buy more stocks. But don’t be fooled: the market has not broadened; it has just found new ways to buy Nvidia.

Everything is turning into AI stocks. This isn’t necessarily bearish, but investors are deluding themselves by calling it "broadening," as if that means diversifying away from AI. It does not. Buying "AI adjacent stocks" and calling it broadening is like ordering a Diet Coke with a Double-Double burger at In-N-Out. Let’s be clear: you still bought a cheeseburger.

Among the big tech companies, who has the least reliance on AI? Apple. Apple's stock price has risen 60% over the past year, recently surpassing Nvidia to become the world's most valuable company again. Amazon, still related to AI but more diversified than other hyperscale cloud providers, has risen 11% over the past year. Microsoft, the core battleground for AI, has fallen 23%.

If I could go long on a basket of stocks, it would be GLP-1. If I could go short on one, it would be AI. But to be clear: I’m not telling you to hold gold bars or cash. I’m always in the market------you never know how fast or irrational it will run. But you should understand how large the market's real exposure to a sector is.

I am a staunch fan of index funds and passive investing: put the money in and let the market do the work. But now we must ask what true diversification actually means. Putting money into the S&P 500 no longer does the job, which means you have to start doing some homework.

The question is: can you find sectors that are truly away from AI?

I would point to one sector: healthcare. This was one of my picks at the beginning of the year, and I stand by it. AI has not yet touched it------which means real returns may still be ahead. But finding these sectors is the challenge investors face now.

Netflix's participation declines, competitive pressure increases

Netflix reported disappointing second-quarter earnings. Revenue grew by 13%, below expectations, and the streaming giant released weak participation data, subsequently announcing it would reduce the frequency of releasing participation metrics, which unsettled investors. The stock price dropped by as much as 8% on Friday.

Netflix has boasted about its transparency; now, that statement seems quite ironic. In the first quarter of 2025, Netflix will stop reporting quarterly subscriber numbers, telling investors to focus on participation. Last week, the company decided to reduce the frequency of its What We Watched participation report from twice a year to once a year starting in 2027.

The last semi-annual participation report looked weak. Total viewing time grew by only 2%, while the subscriber base is estimated to have grown by 10%, meaning daily participation per subscriber fell by 8%.

Netflix has been facing increasing competition from short video providers (especially YouTube). In response, it has added "Clips," a TikTok-style scrolling feature that surfaces short content from its own library, struck video podcast deals with Spotify and Barstool, and reached new licensing agreements with external publishers (BuzzFeed, Condé Nast) to bring new short video content to the platform.

Netflix has lost over $250 billion in market value over the past year, while fellow streaming giant Disney has lost nearly $50 billion. Both are well-managed companies, with revenue and subscriber growth, and prices rising------yet they are being punished for it. This raises an important question: Is streaming just a bad business? Or have Netflix and Disney run out of creativity? Let us know your thoughts in the comments.

In the next six months, OpenAI will acquire enterprise AI company Sierra and appoint Bret Taylor as CEO. Sam Altman will be promoted to chairman. Altman is an innovator, not an operator, while Bret Taylor may be the best enterprise software operator of his generation.