I continue to believe that the broadening out of the “winner” in AI is one of the most interesting narratives from this year. I am also still open to the possibility that NVDA is not the obvious winner when all the dust settles……

AI Civil War: How Google Just Broke the ‘Nvidia Wins Everything’ Narrative

The launch of Gemini 3 and Google’s TPU‑only training run didn’t just add another model to the AI arms race – it shattered the illusion that this cycle would be a clean, winner‑take‑all trade where you just buy $NVDA and $MSFT and go to sleep. We now have two full AI “blocs”: the OpenAI/Nvidia/Oracle ecosystem on one side and the Google/TPU/Broadcom complex on the other, with Meta already flirting with both. That alone is a regime shift. For two years the market treated AI like an organic Nvidia monopoly – 90% share, 75% gross margins, and a CUDA software moat that made alternatives feel theoretical. Now you have a state‑of‑the‑art model trained entirely on Google silicon, Apple and Anthropic already using TPUs, and reports that Meta is in talks to buy them for a dedicated AI data center. That’s not the end of Nvidia’s dominance, but it is the start of real competition in the hottest capex category on earth.

The immediate implication is multiple compression and narrative fragmentation at the top of the stack. When AI looked like a near‑monopoly for Nvidia and a single model provider, the market was happy to pay “infinite” prices for the perceived winner and ignore the risk that everyone else was overpaying and overbuilding. Now the hyperscalers are openly competing with each other in their most capital‑intensive project ever, all building their own chips, all chasing the same customers, and all layering vendor‑financed, SPV‑financed, and balance‑sheet debt on top of it. That doesn’t kill the AI trade – the demand is very real – but it does change the character of it. Hardware and infra in competitive markets historically don’t sustain 75% gross margins forever; over time competition, custom silicon, and purchasing power push pricing and returns back toward earth. Think of Cisco and Intel in the early 2000s: great businesses, but the equity story peaked years before the fundamentals did.

So who wins from an “AI civil war”? In the near term, Google and Broadcom are obvious winners. Alphabet just vaulted back to the center of the AI conversation with Gemini 3, reclaimed narrative leadership after a year of being seen as “behind,” and put a real second source on the table for hyperscalers desperate to diversify away from Nvidia dependency. That alone justifies some re‑rating. Broadcom wins as the quiet arms dealer: it already co‑designs TPUs and other custom AI ASICs, and a world where big platforms split their spend between $NVDA and in‑house accelerators is a world where AVGO sells more connectivity, more custom silicon, and more content per rack. Meta also belongs on the winner list: if it can source competitive TPUs from Google, it gets negotiating leverage with Nvidia, more control over its own AI roadmap, and potentially better economics over a 5–10 year horizon. Down the stack, anyone selling critical “picks and shovels” that are vendor‑agnostic – power, cooling, interconnect, advanced packaging – benefits from a broader, more competitive hardware ecosystem.

On the loser side, the first and most obvious is the “Nvidia will own everything forever” trade, not Nvidia the company. The business is still an absolute monster: Blackwell is selling out, data‑center revenue is exploding, and CUDA is still the deepest software moat in the space. But the equity market was pricing something close to permanent, frictionless dominance: 90%+ share, structurally ultra‑high margins, and no credible second source. Gemini 3 on TPUs, Meta kicking the tires on Google silicon, Anthropic and others spreading workloads across $NVDA, TPUs, and Amazon’s Trainium – that’s how chip cycles always evolve once they get big enough. Over time that means more pricing pressure, more bespoke deals, more “circular” financing that is basically discounting in disguise, and a P/E that has to live with competition instead of pure monopoly narrative. Oracle also slides into the “higher‑risk” bucket here: it’s turning itself into a leveraged AI infra landlord heavily tied to OpenAI, just as customers are demonstrating they want multi‑vendor, multi‑chip strategies and just as Google’s block starts to look like a credible alternative.

The deeper implication for investors is that the AI trade is shifting from a clean, top‑down “own the one winner” to a messy, competitive, capital‑intensive cycle where you get paid more for owning the ecosystem than trying to crown a single king. The more Google’s TPU stack succeeds, the more it forces everyone – Amazon, Microsoft, Meta, even OpenAI’s partners – to diversify their hardware and model dependencies. That’s bad for max‑multiple dreams at the very top, but good for second‑layer beneficiaries: cloud‑agnostic data‑center REITs and infra, power and grid enablers, high‑end networking, timing, packaging, and the software and vertical applications that sit on top of whatever chips win. The “AI civil war” doesn’t end the supercycle; it just means you should stop thinking like a venture capitalist trying to pick the one trillion‑dollar winner and start thinking like a hedge fund manager in a real industry cycle: lean into competition, multiple compression, and second‑order winners, and be very careful about paying bubble prices for any company whose business model quietly depends on nobody else ever catching up.

News vs. Noise: When Japan Sneezes and Crypto Catches Pneumonia

Yesterday was a reminder that the plumbing still runs this market. The first real news was Japan. BoJ Governor Ueda basically put a December hike on the table, JGBs sold off hard, and that move bled straight into Bunds and Treasuries (10Y UST +7 bps, Bund +6 bps, Italy +7 bps, S&P -0.5%). The last time we saw this movie – August 2024 – a BoJ surprise plus weak U.S. data sparked a violent unwind of yen carry and a broad risk puke. This time, positioning is much cleaner and the scale of carry is smaller, so I’m not in “here comes August 2.0” mode. But directionally it’s the same story: Japan is slowly killing the “free yen funding” regime and pushing the global cost of capital higher at the margin. Layer on Trump signaling he’s already picked Powell’s successor (betting markets say Hassett, i.e., dovish) and you get a steeper curve for the wrong reason – not growth, but fears that a politically friendly Fed will cut too far, stoke inflation, and erode what’s left of central bank credibility. The market is trying to price both: a BoJ that’s finally waking up and a Fed that might be leaning easier than the data justify. Net‑net, that still lines up with my medium‑term view: a structurally steepening 2s/10s, the Fed easing into a not‑terrible economy, and an eventually supportive backdrop for risk – but with more rate volatility and more macro landmines along the way.

The second real news was in crypto, and it wasn’t “BTC down 8%” – it was how and who. Bitcoin sold off hard right as Ueda teed up a hike, reminding everyone that BTC now trades like a high‑beta macro asset, not an uncorrelated hedge. Then MicroStrategy’s CEO basically took a sledgehammer to the “Bitcoin forever, never sell” story: talking openly about “Bitcoin winter,” admitting they’d have to dump coins if mNAV breaks 1 and capital markets shut, and forcing the company to scramble out a USD cash‑reserve announcement to calm nerves. For a so‑called “Bitcoin treasury” that’s bought ~460,000 BTC at ~$92k–100k and still trades at a premium to underlying coins you can buy in an ETF, that’s not “smart financial engineering” – it’s a leveraged wrapper with bad entry prices and real forced‑seller risk. That matters because retail in crypto and high‑beta tech is the same cohort; they got margin‑called in October, they’re still rebuilding, and every fresh shock – whether it’s BoJ, DATs, or regulation – tightens the liquidity screws. So the signal isn’t “crypto is dead”; it’s that regime shifts in Japan and cracks in the digital‑asset‑treasury model both raise fragility across the risk complex. BTC itself will find a floor; the tourists in over‑levered wrappers may not.

Takeaways:

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.