Nvidia remains under pressure after entering a technical correction, having fallen 14% from its highest level this year. The AI chip giant now faces a crucial test as earnings from major technology companies over the next two weeks could either reignite investor optimism or trigger another wave of selling.

Rising Competition and Client Risks

Nvidia faces significant risks from the advancement of Chinese AI models, evidenced by Moonshot’s Kimi K3 model beating OpenAI’s ChatGPT and Anthropic’s Claude. Nvidia has invested $10 billion in Anthropic and $30 billion in OpenAI. The growth of Chinese models threatens the market share of these companies.

Simultaneously, Nvidia is encountering strong competition from its own clients. Alphabet is ramping up its TPU sales, with Morgan Stanley expecting the business to generate $13 billion by 2027. OpenAI has unveiled its first chip, manufactured by Broadcom. Microsoft, Amazon, and Meta Platforms are also developing their own inference chips, while competitors like Cerebras and AMD are gaining market share.

A critical risk is that Nvidia’s biggest clients may announce a slowdown in capital expenditure due to rising costs. Companies like Microsoft and Amazon could see their stock prices rise if they declare reduced spending.

Analyst Projections and Valuation

Despite these risks, analysts remain optimistic about Nvidia’s near-term growth. The average estimate for second-quarter revenue is $91 billion, representing a 96% increase year-over-year. For the full year, analysts expect revenue to reach $393 billion.

Nvidia’s valuation has become more attractive amid recent fears. The company’s forward price-to-earnings ratio stands at 22, lower than its five-year average of 43. Additionally, the company maintains a Rule-of-40 metric of over 120%.

Benzinga data indicates the average NVDA stock forecast is $309, significantly higher than the current price of $202. Keybanc recently raised its target from $310 to $330, while DA Davidson maintained a $300 rating. The most optimistic outlook comes from Baird’s Tristan Gerra, with a target of $500.

David Sacks and Bill Ackman warned that China's Kimi K3 model has narrowed America's lead in artificial intelligence, triggering a slide in chip stocks as investors reassessed the compute requirements for the AI race. The alarm followed the release of Kimi K3 from Chinese startup Moonshot, a model that reportedly topped the Frontend Code Arena leaderboard, edging past Anthropic's Claude Fable 5. Sacks, who chairs the President's Council of Advisers on Science and Technology, called the ranking "concerning," citing domestic policy risks such as moves to block new data centers and proposals to make regulators pre-approve frontier models. Ackman replied with one word: "Agreed."

Chip Stocks React

Nvidia Corp. slipped 2% Friday, extending a rough week for semiconductors as traders reassessed how much compute the AI race really needs. The move echoes early 2025's "DeepSeek moment," when a cheap Chinese model briefly erased chip valuations before buyers returned. Micron Technology Inc. has fallen harder this month. The memory maker slid into a bear market on separate worries tied to a rival Chinese IPO and possible export limits, and it rebounded slightly Friday even as the Kimi headlines spread.

Margin Math and Analyst Views

A sharper investor case came from Gavin Baker, chief investment officer at Atreides Management. He argued Kimi K3 could be "net positive for essentially every other company" while squeezing the labs that reportedly enjoy roughly 90% inference margins. Chamath Palihapitiya amplified the point, arguing frontier-grade output has collapsed in price and the model makers' margins can't survive it. "The math ain't mathing," he wrote. That cheap-token framing is contested. Kimi K3 is verbose and reasoning-heavy, so it can burn far more tokens to finish a task, which Baker said makes it 50% to 70% more expensive to run than a comparable US model despite lower headline pricing.

Prediction Market Outlook

On Polymarket, the contract on which company holds the best AI model at the end of the year still gives Anthropic more than 67%, with Moonshot in the low single digits. For a listed way to trade the theme, the cleaner read may be SpaceX, whose Grok family makes it a rare public frontier model maker. Anthropic has reportedly filed to list as soon as this year, while OpenAI is said to have pushed its debut to 2027.