Meta Smart Glasses, Marvell Buy Rating, MACOM Doubles

Several mid-sized AI hardware and platform stories moved today, with Meta’s Connect conference, a new Marvell analyst initiation, and a lesser-known optical chip stock drawing attention from investors tracking the broader AI build-out.

Key points

• Meta announced new privacy-focused smart glasses without cameras at its Connect 2026 conference; GF Value pegs the stock at roughly 12.5% undervalued.
• Seaport Research Partners initiated coverage of Marvell Technology (NASDAQ: MRVL) with a Buy rating and a $270 price target, though GF Value flags the stock as 113% overvalued by that metric.
• MACOM Technology Solutions (NASDAQ: MTSI) announced a 3.2-terabit optical chipset built on eight 448-Gbps PAM-4 channels; the stock has more than doubled over the past year.
• Qnity Electronics (NYSE: Q) joined the PHLX Semiconductor Sector Index on September 21, a milestone for the semiconductor materials specialist.
• Andreessen Horowitz launched the Horowitz Andreessen Academy, a full-time San Francisco tech school for high school graduates, with Nvidia, OpenAI, Meta, and Anthropic as founding partners.

What is Meta doing with glasses that have no cameras?

The camera-free design is a deliberate pivot. Meta’s earlier Ray-Ban smart glasses drew scrutiny over covert photography risks, and the new models announced at Connect 2026 appear to address that head-on. Dropping the camera removes a key regulatory friction point in Europe and opens the product to venues, workplaces, and contexts where camera wearables are prohibited.

For investors, the question is whether privacy-by-design expands the addressable market enough to matter. The glasses segment is still small relative to Meta’s core advertising business, but the company has been consistent in funding this hardware line over multiple product cycles. GurufFocus’s GF Value model currently puts META at roughly 12.5% undervalued, though valuation models of this type reflect backward-looking fundamentals more than hardware optionality. Nothing here is a buy or sell signal.

Marvell: a Buy rating meets a bearish valuation screen

Seaport Research’s initiation of Marvell with a $270 price target lands on the same day GF Value flags the stock as 113% overvalued. That gap is worth understanding before drawing any conclusions. GF Value is a formula-driven intrinsic value estimate; Seaport’s Buy is presumably built on a forward earnings model that weights Marvell’s custom AI silicon pipeline and data center interconnect exposure more heavily than trailing metrics would suggest.

Marvell has been one of the more discussed custom ASIC plays as hyperscalers look to reduce Nvidia dependence for specific workloads. A $270 target from a new covering analyst adds to the bull case narrative, but the wide divergence between the two valuation signals is a reminder that MRVL is a stock where assumptions about AI infrastructure spending timelines drive outcomes more than current financials do.

MACOM and Qnity: the optical and materials layer gets attention

MACOM’s 3.2-terabit optical chipset, announced September 17 and drawing fresh coverage today, targets the interconnect bottleneck inside AI data centers. As GPU clusters scale, the bandwidth between compute nodes becomes a constraint. MACOM’s eight-channel 448-Gbps PAM-4 design pushes at that limit. The stock has already doubled over the past year, so investors considering MTSI now are buying into a move that has largely happened rather than one that is just starting.

Separately, Qnity Electronics joining the PHLX Semiconductor Index is a structural development. Index inclusion typically brings passive fund buying and higher trading liquidity, which can be a short-term price catalyst independent of the company’s underlying fundamentals. GurufFocus notes mixed valuation signals on Qnity, so the index effect is the cleaner near-term story here.

A16z’s academy: what it signals for AI talent supply

The Horowitz Andreessen Academy is a non-financial story with financial underpinnings. Andreessen Horowitz’s portfolio depends on a pipeline of AI-literate builders. Pulling Nvidia, OpenAI, Meta, and Anthropic in as founding partners turns the academy into a de facto recruiting and curriculum-shaping vehicle for the firms that matter most to a16z’s fund returns. For investors, the more interesting read is that these companies are willing to fund alternative credential pipelines, suggesting they see a real shortage of AI-ready talent at the entry level that traditional universities are not filling fast enough.

Today’s session was relatively light on hard market-moving data. The Alibaba chip and data center story covered yesterday remains the week’s biggest single-company development in AI infrastructure.