Nvidia's $91B Target, Meta's Chip Timeline, Intel Backed

Three separate AI infrastructure stories moved in parallel today: Nvidia’s forward revenue case got a fresh look, Meta locked in a chip production timeline, and the Trump administration stepped in to support Intel’s comeback bid.

Key points

• Nvidia’s $91B revenue target is reaffirmed by analysts, with fears over a Kyber architecture delay described as unconfirmed.
• Meta plans to begin manufacturing its own AI chip in September, targeting 14 gigawatts of total compute capacity by next year.
• The Trump administration is actively backing Intel’s turnaround by pushing potential customers toward the company.
• Micron shares rose 7% after announcing a new round of U.S. chipmaking investments.
• xAI’s Grok 4.5 launched at $2 per million input tokens, well below flagship rates from Anthropic and OpenAI, trained on tens of thousands of Nvidia GB300 GPUs.

Does the market still have Nvidia wrong?

A fresh Seeking Alpha piece argues that the market continues to underestimate Nvidia’s next phase. The central claim is that concerns about a delay to the Kyber architecture remain unconfirmed, and the $91B revenue target analysts have modeled is still intact.

This is a meaningful distinction for investors tracking the stock. If Kyber slips, it would affect the cadence of Nvidia’s product cycle and, by extension, the timing of hyperscaler upgrade spending. But the piece is careful to frame those fears as unsubstantiated at this point rather than settled. Nvidia has repeatedly managed supply transitions without the demand destruction that skeptics predicted, so the burden of proof on a delay story is reasonably high.

The confirmation that Grok 4.5 was co-trained using tens of thousands of Nvidia GB300 GPUs adds one more data point to the demand picture. Even as xAI competes aggressively on model pricing, it is spending heavily on Nvidia hardware to get there.

Meta’s chip timeline gets specific

Meta’s AI chip ambitions are not new, but a September production start date is. An internal memo cited by The Hindu shows the company planning to manufacture its own AI chip starting that month, as part of a broader push to reach 14 gigawatts of computing capacity in 2027.

This follows the earlier reporting on Meta’s Iris chip that we covered earlier this week. The new detail here is the production schedule and the scale target. Fourteen gigawatts is a substantial number, and reaching it by next year implies aggressive procurement and buildout across both proprietary and third-party hardware. Meta has not said the in-house chip will replace Nvidia’s GPUs wholesale. For now, both paths appear to be running in parallel.

Intel gets political cover. Does it help?

The Trump administration is reportedly pushing potential customers toward Intel as part of a broader effort to support the company’s manufacturing recovery. The sources on this story are thin, limited to brief dispatches without detailed attribution, so the precise mechanics of that push are unclear.

What is clear is that Intel’s turnaround remains politically important to U.S. semiconductor policy. The company is central to the domestic fabrication capacity that CHIPS Act funding was designed to build. Government encouragement of customer relationships is not the same as a technical or competitive breakthrough, but it signals continued institutional support that could matter for large procurement decisions.

Micron and the broader chip trade

Micron’s 7% single-day gain following its U.S. investment announcement is worth flagging in context. A separate report notes that an overlooked chip ETF is outpacing the biggest AI names in 2026, with memory and equipment makers delivering triple-digit gains that dwarf the broader market. The fund generating most of that upside manages a fraction of the assets held by the sector’s largest ETFs.

That pattern, smaller and more specialized vehicles capturing the bulk of returns that broader funds dilute, is a recurring feature of the current semiconductor cycle. Investors tracking chip exposure through large-cap-heavy products may be getting less of the actual outperformance than the headline numbers suggest.

Nothing on this site constitutes investment advice. This update is for informational purposes only.