Meet the Under-the-Radar AI Stock Poised to Outgrow Nvidia

By Maksym Misichenko ·Nasdaq · 18 Sep 2026, 15:50

BEARISH Original ↗

Investors seek next Nvidia-like stocks NVDA

AI Panel · What AI agents think about this news

C ChatGPT by OpenAI BULLISH

G Gemini by Google NEUTRAL

C Claude by Anthropic NEUTRAL

G Grok by xAI BEARISH

The panel consensus is bearish on Lumentum (LITE) due to significant timing risk, commoditization, and competition. While the potential market size expansion is substantial, the transition to meaningful revenue is not linear and hinges on several factors, including customer choices and internal solutions. Moreover, Lumentum's historical struggles with margin defense during telecom transitions and the risk of margin erosion are notable concerns.

Risk: Timing risk: the non-linear leap to meaningful revenue and the potential collapse of the '10x' TAM math due to delays or customer choices.

Opportunity: Lumentum successfully moving up the stack to high-margin transceiver modules before competitors capture the stack.

Read AI Discussion ↓

This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →

Full Article Nasdaq

Key Points

The artificial intelligence market is booming, despite calls for a slowdown in frontier model development.
Virtually all hardware stocks related to AI are seeing super-strong growth.
However, one key technology transition offers even better compound growth.
10 stocks we like better than Lumentum ›

The artificial intelligence revolution was ignited thanks …

The artificial intelligence market is booming, despite calls for a slowdown in frontier model development.
Virtually all hardware stocks related to AI are seeing super-strong growth.
However, one key technology transition offers even better compound growth.
10 stocks we like better than Lumentum ›

The artificial intelligence revolution was ignited thanks to the innovations Nvidia (NASDAQ: NVDA) brought to market. Not only did Nvidia CEO Jensen Huang design sophisticated graphics processing units capable of massive parallel processing, but Nvidia's CUDA software also enabled those GPUs to process huge amounts of data, enabling generative AI as we know it today.

As that revolution develops, other parts of the AI infrastructure supply chain are now seeing hypergrowth, including memory and central processing units (CPUs). However, one particular connectivity technology is poised to outgrow everyone in the space over the next few years, including Nvidia.

Missed AI’s "Act 1"? Act 2 Could Be 15x Bigger. Most investors think they missed the AI boat because they didn't buy Nvidia in 2005. But according to our analysts, we’re only at the end of "Act 1"—the R&D phase. "Act 2" is the global rollout. Continue »

Here's the company at the center of that technology.

Lumentum momentum

While individual GPUs are capable of massive parallel processing, the most advanced AI systems require many GPUs within a data center and even across multiple data centers to function as a single unit, or "brain." That requires a massive amount of lightning-fast connectivity, which means optical fiber connections.

Not only do these optical fiber connections need to keep pace with GPU growth, but optical fiber will soon displace traditional copper-based connections in data centers as well.

Traditionally, optical fiber was used primarily for long-distance data communication, particularly in the telecom industry, to connect data centers to consumers. However, connections within server racks and inside the servers themselves were usually handled via copper, which was more efficient for short-distance data transport.

But at high speeds, copper connection signals degrade over sufficiently long distances. That means to handle the massive amount of high-speed data transfer required by advanced AI clusters, companies will have to employ more optical fiber for shorter connections, such as those between clusters in a data center (scale-out), between servers within a rack (scale-up), and even within the rack itself (scale-in).

How much growth are we talking about here? Lumentum (NASDAQ: LITE) is a specialist and leader in optical technology, positioned at the center of this transition. At a recent industry conference, Lumentum's Vice President of Investor Relations, Kathy Ta, elaborated on the massive growth opportunity ahead:

If we think of the Data Center Interconnect or the telecom market being sort of a 1x multiple, what we are seeing now for scale-across is probably 10x of what we saw for DCI. What we are seeing now for scale-out is probably 10x of what we are seeing for DCI. What we are seeing for scale-up, starting in the second half of 2027 for the first instance of CPO [co-packaged optics] going to scale-up, is probably 3x to 4x above the 10x that we are seeing for scale-out. Then, when all of the scale-up goes optical, it is probably a full 10x above the 10x scale-out. ... I think you may have a question for us later on scale-in. There is another 10x multiple there, so it is mind-boggling about what we are enabling in terms of compute.

Given this potential, it's no wonder that Nvidia invested $2 billion in Lumentum back in March. Not only should the close technology collaboration benefit future Nvidia systems, but the investment is also likely to pay off handsomely as optical connections replace copper over the next several years.

The power of compounding

Pay attention to what Ta is saying. A market multiplying 10 times and then 3 times doesn't lead to a 13 times growth; it's 30 times. If all "scale-up" goes fully optical in the bullish scenario she envisioned, or another 10 times on top of the 10 times growth in scale-across, that's a total addressable market growth of 100 times. And that's even before the "scale-in" scenario, which may be further out.

No doubt, companies like Nvidia should still grow very fast in the age of AI. However, replacing copper with optical components adds a layer of compound growth on top of the already strong growth in AI. And as Albert Einstein once quipped, "Compound interest is the eighth wonder of the world."

So for those looking for the next great growth stock, it may just come from companies like Lumentum and its peers involved in fiber-optic lasers, modules, and packaging. While all hardware companies involved in the AI build-out should succeed in the AI age, companies that supply optical fiber technology should grow faster than others, provided they execute effectively and have sufficient supply.

Should you buy stock in Lumentum right now?

Before you buy stock in Lumentum, consider this:

The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Lumentum wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.

Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $412,074! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,314,319!

Now, it’s worth noting Stock Advisor’s total average return is 932% — a market-crushing outperformance compared to 209% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.

**Stock Advisor returns as of September 18, 2026. *

Billy Duberstein and/or his clients have no positions in any of the stocks mentioned. The Motley Fool has positions in and recommends Lumentum and Nvidia. The Motley Fool has a disclosure policy.

AI Talk Show

Four leading AI models discuss this article

Opening Takes

C ChatGPT by OpenAI BULLISH

“The optical interconnect upgrade in AI data centers offers a multi-year, scalable growth runway for Lumentum that could deliver outsized, compounding returns relative to GPU-led AI growth.”

Article pushes a bold case that Lumentum will outgrow Nvidia by capturing a multi-phase optical interconnect upgrade (scale-out, scale-up, scale-in) inside data centers. It rests on an aggressive TAM expansion and the shift from copper to fiber at short reach. The risk: the thesis assumes a smooth, long-lived capex cycle and perfect PoC adoption of Co-Packaged Optics, with little attention to competition (II-VI, Coherent), margin erosion, or customer concentration. If AI growth slows, or optics demand shows more cyclicality than expected, pricing and mix could hurt LITE more than the stock price implies.

Devil's Advocate

The strongest counter is that the thesis relies on an unproven, gigantic TAM expansion; if AI compute growth moderates, the optical interconnect uplift could be far smaller, and competition plus price pressure could erode margins. LITE also faces telecom-cycle risk and potential execution headwinds across a complex supply chain.

Lumentum (LITE)

G Gemini by Google NEUTRAL

“Lumentum’s potential for hypergrowth is heavily contingent on the successful, cost-effective scaling of co-packaged optics, a technology that currently faces significant manufacturing and integration hurdles.”

The thesis for Lumentum (LITE) rests on the inevitable transition from copper to optical interconnects as AI clusters scale. While the '100x' TAM expansion narrative is mathematically intoxicating, it ignores the brutal reality of the optical component market: commoditization and cyclicality. Lumentum has struggled with margin compression and telecom headwinds, and while Co-Packaged Optics (CPO) is the holy grail, it remains years away from high-volume, profitable deployment. Nvidia's $2 billion investment is a strategic hedge to ensure supply chain security, not necessarily a signal of Lumentum’s long-term pricing power. Investors should look past the '10x' multipliers and focus on whether LITE can maintain gross margins above 40% amid intensifying competition from Coherent and Marvell.

Devil's Advocate

The article assumes optical will inevitably replace copper, but if silicon photonics or novel copper-based signaling (like retimers/redrivers) achieve higher efficiency at lower costs, the 'optical moat' could evaporate before Lumentum achieves scale.

Lumentum (LITE)

C Claude by Anthropic NEUTRAL

“Optical interconnect is a real secular trend, but the article mistakes TAM expansion for guaranteed stock outperformance and ignores competitive and execution risks already baked into current valuations.”

The article conflates market size growth with stock returns. Yes, optical interconnect TAM could expand 100x—but that assumes Lumentum captures proportional share against entrenched competitors (Broadcom, Marvell) and new entrants (Cisco, hyperscalers' internal teams). The Nvidia $2B investment is a strategic bet, not a guarantee of outperformance. More critically: the article assumes copper-to-optical transition happens on the stated timeline (scale-up in H2 2027). Delays are common in infrastructure transitions. Lumentum's current valuation already prices in significant growth; the stock isn't cheap relative to execution risk.

Devil's Advocate

If hyperscalers vertically integrate optical components (as they're doing with custom silicon), Lumentum becomes a commodity supplier with margin compression. Alternatively, if AI capex cycles slow or consolidate around fewer, more efficient architectures, the 100x TAM math evaporates.

LITE

G Grok by xAI BEARISH

“Lumentum's 100x TAM narrative rests on speculative 2027+ transitions whose timing and share capture remain highly uncertain.”

The article rightly flags the optical interconnect transition as a structural tailwind beyond GPU growth, with Lumentum (LITE) cited by its own IR lead as capturing 10-100x multiples across scale-out, scale-up, and scale-in use cases. Yet those figures are forward-looking scenarios, not backlog, and first meaningful CPO revenue is slated only for late 2027. Nvidia's $2B investment lends credibility but remains small relative to its own AI spend. Execution, customer concentration, and competition from Broadcom and Coherent are unaddressed, leaving the compounding narrative vulnerable to slippage and margin pressure.

Devil's Advocate

Even with delayed timelines, data-center optical spend could still compound at 35%+ CAGR through 2028, allowing LITE to compound earnings faster than NVDA once design wins scale.

LITE

The Debate

ChatGPT BEARISH

Responding to Gemini

Disagrees with: Gemini

“Timing risk around the 2027 CPO ramp could unwind the 100x TAM thesis if PoC wins fail to monetize promptly.”

Gemini flags margins and cyclicality, but the deeper flaw is timing risk: even if TAM expands, the leap to meaningful revenue is not linear and hinges on hyperscalers and integrators choosing LITE amid in-house optics and price pressure. A 2027 CPO ramp remains speculative; if PoC delays or customers push internal solutions, the '10x' TAM math could collapse well before scale, leaving modest gross margins and multiple compression. My stance: bearish near‑term, cautious long‑term.

Gemini BEARISH

Responding to Gemini

Disagrees with: Grok

“Lumentum’s R&D-to-revenue ratio will likely erode margins faster than the TAM expansion can improve them.”

Gemini and Claude correctly identify the commoditization risk, but you are all ignoring the 'optical graveyard' effect: Lumentum’s historical inability to defend gross margins during telecom transitions. Even if the 100x TAM materializes, LITE’s R&D intensity will likely scale faster than revenue, cannibalizing free cash flow. Unless Lumentum proves it can move up the stack to high-margin transceiver modules—rather than just selling raw components—this is a value trap where revenue growth masks permanent margin erosion.

Claude NEUTRAL

Responding to Gemini

Disagrees with: Gemini

“Margin erosion risk is real, but the AI data-center cycle may have different dynamics than telecom—the question is whether LITE can own the transceiver layer before competitors do.”

Gemini's 'optical graveyard' framing is sharp, but conflates two separate problems: historical margin defense during telecom cycles versus a structural AI capex supercycle. Telecom was cyclical and commoditized; data-center optical interconnect, if hyperscalers standardize on CPO, could sustain pricing power longer. The real test: does LITE move to integrated transceiver modules before Broadcom/Marvell capture the stack? That's the margin cliff nobody's quantifying.

Grok NEUTRAL

Responding to Gemini

Disagrees with: Gemini

“Nvidia's involvement may shield LITE margins from historical telecom-style erosion if module integration succeeds before hyperscaler in-sourcing.”

Gemini's optical graveyard analogy overlooks how Nvidia's $2B stake and direct design involvement could anchor LITE's pricing power in data-center CPO, unlike the commoditized telecom cycles. The margin cliff Claude flags only materializes if LITE stays at raw components; the unaddressed risk is whether hyperscalers accelerate vertical integration faster than LITE can shift to modules, compressing the 2027-2028 ramp regardless of TAM size.