Everyone wants to debate whether AI is a bubble. Fine. Let them. The better question is simpler and more profitable: where does the money have to go, even if the hype cools? And right now, Big Tech is telling you—loudly—where it’s going: into the physical buildout. The world’s largest tech companies are lining up hundreds of billions in 2026 capex for AI infrastructure. The WSJ puts the combined total for Microsoft, Meta, Amazon, and Alphabet at up to ~$670B—a historically massive private capex wave. And Alphabet is talking about $175–$185B of capex in 2026 alone as it pushes AI + cloud capacity.
Here’s the catch: you can buy all the GPUs you want… but if you can’t move data fast enough between them, the factory stalls. That’s the unsexy truth of AI: the bottleneck isn’t always compute. It’s throughput.

AI training and inference aren’t “one big chip.” They’re clusters—tens of thousands of GPUs/accelerators acting like one machine. That machine lives or dies on how fast it can shuffle data across racks, rows, and buildings. And at the bleeding edge (800G → 1.6T and beyond), “just use copper” stops working. Loss, heat, power draw, distance limits… physics shows up like a tax collector.

That’s why Nvidia and the networking ecosystem are pushing toward silicon photonics and co-packaged optics (CPO) as the next architecture shift—moving optical engines closer to the switch ASIC to reduce loss and power. Nvidia has explicitly laid out roadmaps pointing to photonics/CPO becoming essential for next-gen AI data center networks around 2026.

And it’s not just CPO. There’s also a second quiet revolution happening in parallel: optical circuit switching (OCS)—a way to reconfigure AI clusters with far lower power overhead than brute-force packet switching, increasingly discussed as AI connectivity demands explode. The Open Compute Project even created a dedicated OCS subproject to standardize and accelerate this category.

Translation: the “pipes” are becoming the product.

Think of optical/interconnect as a three-wave buildout. Each wave has different time horizons and different winners.

This is the “no excuses” part. If hyperscalers are building clusters, they need transceivers and optical module supply chains. And we’re already seeing companies talk about robust data center demand tied to 800G and 1.6T products.

Wave 1 is what I call the “capex-to-shipments translation layer.” It’s where PowerPoint turns into purchase orders.

OCS matters because AI clusters increasingly need flexible topologies: different training jobs, different traffic patterns, different bottlenecks. If you can reconfigure cheaply and fast, you reduce “wasted” compute time.

Lumentum has publicly announced an OCS platform (R64) targeted for AI data centers, with customer sampling beginning in calendar Q4 2025 and general availability expected in the second half of 2026.

Separately, industry research cited by Fibre Systems (referencing Cignal AI) has forecast the OCS market reaching at least $2.5B by 2029, with the forecast revised upward versus earlier expectations.

Wave 2 is where “AI networking” stops being just Ethernet switches and becomes architectural advantage.

CPO is the long game—and it’s the most misunderstood. Investors hear “CPO” and think, “Oh, that kills pluggable optics.” Maybe eventually. But in reality, CPO is a multi-year transition, and it expands the photonics ecosystem even as it changes the form factor.

You can see the strategic positioning happening already:

Wave 3 is the “platform war” layer. It’s where the architecture choices reshape the supply chain.

Not “recommendations.” Just the map.

1) Optical component + photonics manufacturers (the toll booths)
These are the businesses sitting closest to the conversion of AI capex into shipped optical hardware—lasers, photonics components, modules, and (in some cases) OCS adjacency.

2) Contract manufacturing / “arms dealer” enablers
If you want exposure to the ramp without betting on a single OEM brand, you often go to the people who build the thing.

3) Silicon photonics + networking platform ecosystems (the architects)
The platform winners monetize the shift: switch ASICs, NICs, photonics platforms, co-packaging infrastructure.

Examples to frame the category: Nvidia, Broadcom, Marvell, AMD (each making moves around photonics / CPO / optical fabric).

4) “Boring but mandatory” picks-and-shovels adjacent to optics
As embodied AI and data centers scale, they pull forward demand for:

(These don’t need AI hype. They need volume.)

1) Anything levered to “AI compute is infinite and frictionless”
When networking becomes the choke point, software narratives that assume endless cheap compute can get repriced—fast. Even the market’s recent tone suggests investors are becoming more skeptical of spend without ROI.

2) Overcrowded optics momentum trades
This is the cleanest bottleneck trade… which means it can also become the most crowded, the fastest. When optics gets “hot,” it can whip around violently. (The plumbing trade isn’t immune to positioning.)

3) One-customer, one-form-factor fragility
As architectures shift (pluggable → CPO, or fixed topologies → OCS-enabled), vendors too dependent on one hyperscaler, one module type, or one generation can get blindsided.

A real example of this tension: investor fears have periodically flared that Nvidia’s CPO roadmap could reduce demand for traditional optical transceivers—some analysts have argued those fears are overblown, but the fact that the debate exists tells you where the risk premium is rising.

If you want to run this theme like a pro, ignore the memes and watch these:

“Advanced tech” narratives almost always converge on sensors, signals, and information advantage. And that is photonics.

AARO has explicitly explained that UAP imagery can remain classified not because the object is exotic, but because raw imagery and metadata can reveal sensitive sensor/platform capabilities (resolution, metadata, limitations). That’s optics, sensor fusion, detection, tracking, and—yes—information control.

At the same time, AARO and DoD have repeatedly stated they’ve found no verifiable evidence that UAP cases represent extraterrestrial activity or that the U.S. has access to extraterrestrial tech.

And then you have the “narrative accelerant”: Trump publicly claiming a secret pulsed-energy “Discombobulator” capability was used in the Venezuela operation to make equipment “not work.” That is a claim—not a technical disclosure—but it reinforces the same direction of travel: modern conflict increasingly revolves around non-kinetic, invisible effects (electronics, sensors, comms).

So whether you’re watching AI data centers or “UFO disclosure risk,” the investable common denominator is the same: the infrastructure that moves, senses, and controls information. In markets, that shows up as optical interconnect.

The AI boom is evolving from “who has the best model?” to “who can build the biggest machine?”
And the biggest machine has a hard limit: how fast it can move data.

That’s why optical/interconnect is the cleanest throughput bottleneck trade on the board:

If the market keeps rotating away from story-stocks and toward “real economy” constraints, light-based infrastructure is where the next leg gets built.

If you only read the headlines, January looked like a clean win.

Payrolls came in at +130,000 versus a consensus +55,000. Unemployment ticked down to 4.3%. Politicians immediately took a victory lap.

That’s the noise.

The news is what happened in the background… and what it implies for rates, multiples, and why AI is now knocking over “boring” business models one industry at a time.

Buried inside the release was a benchmark revision that quietly re-writes the last year of “strength.”

Translation: January’s headline number is a snapshot. The revision is the trend.

And the trend says: the labor market isn’t “breaking,” but it’s not remotely as tight as the narrative has been priced for.

Here’s the simplest way to frame labor conditions right now:

That’s roughly a 900,000-person gap where more people are looking for work than there are jobs available.

Historically, that’s not a “soft landing” victory lap indicator. It’s a late-cycle warning light—even if it doesn’t guarantee recession on its own.

Stack that on top of what’s been trickling in elsewhere:

It’s a labor market that’s cooling… and it’s cooling in a way that can be masked for months by headline payroll noise.

Investors spent the early part of this year trying to “will” rate cuts into existence.

But the Fed doesn’t cut because you want your portfolio to feel better.
The Fed cuts when it has to.

And if you’re looking at:

…then the Fed has zero urgency.

The market’s implied probability of a March cut has collapsed to “basically nothing,” and Fed speakers have already started leaning into “hold steady for longer.”

The bigger risk isn’t just “higher for longer” policy rates. It’s the balance sheet.
If the next regime (Warsh or otherwise) pairs any easing with balance-sheet contraction, you can get the worst cocktail for risk assets: rate relief headlines with liquidity tightening underneath.

That’s how you end up with a market that looks calm at the index level… while single names get vaporized.

Here’s what’s happening in plain English:

We are in a market where the index can hover near highs… while entire pockets of stocks get sent to the shadow realm.

Why?

Because investors aren’t repricing “growth.”
They’re repricing certainty.

For the last decade, Wall Street paid premium multiples for anything that looked like:

In 2026, those same qualities are now being interpreted differently:

So the market is doing what it always does in a transition:

It’s not waiting for disruption to fully arrive.
It’s repricing the risk premium today.

Last week it was software and business analytics—anything “recurring revenue” with a premium multiple.

Then it was insurance brokers—the distribution layer that collects fees for matchmaking buyers and policies.

The other day it was wealth management / broker platforms—the business model built on being the interface, the relationship, the gatekeeper.

This is the new pattern:

That’s why the selling looks “irrational” in the moment.

It’s not about next quarter’s earnings.
It’s about what multiple the market wants to pay for “certainty” when the world suddenly feels less certain.

Now add gasoline to the fire: positioning.

After the factor unwind and the violent rotations, a lot of books look like this:

When positioning gets that one-sided, you don’t need good news for a snapback.
You just need the selling pressure to stop.

That’s why you can get a market where:

So yes — the rotation out of software and the Mag 7 can snap back hard, even if the longer-term regime is still broadening and dispersion stays elevated.

(If you want the infrastructure angle: this is exactly why the “AI infrastructure” crowd keeps getting pulled back and then re-bid — the bottleneck isn’t demand, it’s power, execution, and speed-to-market.)

The headline payroll number was fine.

The revision was the message: the labor market has been softer than advertised.
The Fed is the implication: don’t assume fast cuts; assume policy uncertainty and liquidity sensitivity.
And AI disruption is the market regime: it’s not one crash—it's a rolling series of “surprise” victims, where valuation gets repriced before fundamentals break.

This is not a “buy the index and go to sleep” year.

This is a year where the index can look calm…
while your portfolio learns what “dispersion” really means.

Viking Therapeutics (VKTX) — the “meme-y” way to play weight loss (and why it’s in MEMY and NOT NVO/LLY)

When people say VKTX just had “blowout earnings,” what they usually mean isn’t “profits” — it’s momentum + milestones. Viking is still in heavy build mode: in its latest update it reported a Q4 net loss of $157.7M (–$1.38/share) as R&D spending ramps, and it exited 2025 with ~$706M in cash/cash equivalents/short-term investments.

Here’s why the tape still treats VKTX like a headline stock: the entire equity story is a probability-weighted bet on VK2735 (their GLP-1/GIP dual agonist) and whether it becomes a real contender in obesity. Viking says its Phase 3 VANQUISH program for injectable VK2735 is underway (with VANQUISH-1 fully enrolled >4,500 patients and VANQUISH-2 expected to complete enrollment in 1Q26). It also plans to advance oral VK2735 into Phase 3 (expected 3Q26), and it has maintenance-dosing data targeted for 3Q26 — which matters because the “end game” in obesity is durability + tolerability + adherence, not just week-13 weight loss.

I’m a huge believer in the weight loss drug theme so why VKTX belongs in “meme-y” (MEMY) and not alongside NVO/LLY when you want exposure?

Because VKTX is “weight-loss optionality,” not “weight-loss cash flow.”

So if you’re building a “weight-loss exposure stack,” think of it like this:

That’s why VKTX is “meme-y”: it’s fundamentally a narrative + catalyst + volatility vehicle. When the market’s in a risk-on, “story-stock” mood, VKTX can outperform dramatically. When the market wants proof (or de-risks biotech), it can get punished even if the long-term thesis hasn’t changed.

Our next webinar…..

Fri, Feb 20, 2PM EST

How to position your portfolio before Washington admits it has non‑human technology, and where the first trillion dollars of “alien alpha” could flow.

You’re not crazy if you believe in UFOs or UAP (unidentified anomalous phenomena). 

In fact, you want to be ready for the day when we’re told, for real, that we’re not alone.

You won’t read about it in the Wall Street Journal or hear about it on CNBC — yet. 

But “Disclosure Day” is coming. 

And the reality of UAP could trigger a shift in global markets that — no hyperbole — makes the Internet boom and the AI explosion seem tiny.

Matt Tuttle, CEO of Tuttle Capital Management ($4 billion AUM), is an ETF rebel who’s spent decades trading big, unexpected market moves.

He created his H.E.A.T. investing framework—Hedges, Edges, Asymmetry, Themes—to turn left-field events into high-conviction opportunities.

Now, he's deploying that exact playbook on the one catalyst almost no one's positioned for...

In a free live “Disclosure Day” briefing, Matt will explain:

✅ Why this isn’t about tinfoil hats. A move from rumor to reality could shift seismic capital across defense, energy, materials, and data – with clear winners and losers.

✅ The “Disclosure Debris:” How small hints can move big money before any big speech from Washington. And why hearings, leaks, and half‑answers already matter more than one big moment if you want a shot at alien alpha.

✅ Early matters – even if you might feel a bit crazy: A simple way to look at UFO news that lets you stay sane, stay skeptical, and still be in position if this really is the next trillion‑dollar theme.

✅ Who could win, who could lose, and when it’s too late: Which sectors of defense, energy, and materials might see money rush in first, and how to think about bet size before everyone on TV is yelling about UFO trades.

✅ The one belief shift that could change how you see every headline about UFOs and tech: The real question isn’t “is this true?” but “what if enough other investors decide it is?”

✅ A 30‑day playbook for the month after confirmation: A practical way to think about reallocating, hedging, and positioning if Washington ever admits more than it already has – without abandoning your own risk limits.

PLUS . . .  you’ll get a free copy of Matt’s Why The UAP Thematic Frontier May Be Closer – And Far Larger – Than You Might Think briefing.

Focus on areas that AI isn’t going to eat……

The H.E.A.T. (Hedge, Edge, Asymmetry and Theme) Formula is designed to empower investors to spot opportunities, think independently, make smarter (often contrarian) moves, and build real wealth.

The views and opinions expressed herein are those of the Chief Executive Officer and Portfolio Manager for Tuttle Capital Management (TCM) and are subject to change without notice. The data and information provided is derived from sources deemed to be reliable but we cannot guarantee its accuracy. Investing in securities is subject to risk including the possible loss of principal. Trade notifications are for informational purposes only. TCM offers fully transparent ETFs and provides trade information for all actively managed ETFs. TCM's statements are not an endorsement of any company or a recommendation to buy, sell or hold any security. Trade notification files are not provided until full trade execution at the end of a trading day. The time stamp of the email is the time of file upload and not necessarily the exact time of the trades. TCM is not a commodity trading advisor and content provided regarding commodity interests is for informational purposes only and should not be construed as a recommendation. Investment recommendations for any securities or product may be made only after a comprehensive suitability review of the investor’s financial situation.© 2025 Tuttle Capital Management, LLC (TCM). TCM is a SEC-Registered Investment Adviser. All rights reserved.