원문 보기
Well, let's get straight to that explanation. Today we are analyzing one of the most significant and, frankly, quite shocking portfolio rotations of 2026. And it's the work of none other than ARK Invest's Katie Wood. If you've been following the markets, you know that ARK isn't shy about making bold moves.
But in about 10 days, we saw a massive tactical rebalancing that literally shook the financial world. We will rely solely on data, looking at specific deals and logic to understand what this means for the market and, more importantly, for your portfolio. Just stop for a second and think about that number.
95 million dollars. That's how many AMD shares, which were soaring, were just abruptly dumped from ARK's portfolio. And to be clear, this was not a gradual reduction of a few stocks to rebalance the weight. Absolutely not. This was a significant, decisive sell-off of shares in a major semiconductor player at the height of the boom in the sector.
When a famous fund manager dumps nearly a hundred million dollars of his beloved tech company, it makes us all stop and take notice. Naturally, seeing a move of this magnitude elicits an instant reaction, right? This forces us to ask the main question. Wait, is Katie Wood backing away from artificial intelligence?
Because selling an AI darling like AMD might seem like a sudden loss of faith if viewed superficially. You start to think: are the grades too high? Is the "AI trade" over? But, as we will now see by examining the transaction logs, this is not a retreat from the sector at all. On the contrary, this is a very calculated aggressive redistribution.
So, let's outline a timeline to better understand the situation. In just 10 days in late August and early September 2026, we saw the unfolding of this calculated trading madness. On August 26, ARK begins the process by selling $18.1 million in AMD stock and acquiring Broadcom. Then, barely 48 hours later, on August 28, — bang, the main rotation takes place .
They are dumping nearly $73 million more of AMD stock and investing over $124 million directly into Nvidia, Broadcom, and Cerebras. And finally, on September 1st, there's this sudden pivot to space infrastructure, with the purchase of over $44 million worth of Rocket Lab stock. Honestly, it's the speed of this repositioning that makes it so exciting.
The mystery deepens even further when we look at the state of AMD's business. So, why sell now? On paper, their performance was simply incredible. They had record revenue for the second quarter of 2026 of $11.54 billion. Data center revenue alone soared by a staggering 107% to $6.7 billion. Additionally, earnings per share easily beat analysts' forecasts.
The company was operating at full capacity, establishing itself as a vital second supplier for hyperscalers desperate for silicon. By all the traditional metrics we use, this was a stock that was definitely worth holding. So with numbers like that, why get rid of it? There are two main “red flags” for a portfolio manager assessing short-term risks.
First, stocks simply rose too much and too fast. As of August, they had grown by approximately 113% since the beginning of the year. They have already accounted for a huge part of future success in the field of AI. In fact, Zacks Research downgraded the stock to a hold rating only because the upside potential had been exhausted.
Secondly, and perhaps more importantly, internal signals began to turn bright red. In August 2026 alone, AMD management, namely the CEO and CFO, sold shares worth almost $49 million. Listen, when insiders withdraw such amounts, the smart money pays attention to it . But if you currently hold AMD in your portfolio, don't panic.
Wood's actions actually tell a rather subtle story. This is profit-taking, not a collapse of the investment thesis. The reality is that Arc still owns about $160 million worth of AMD through its various ETFs. So Wood is not at all against the long-term viability of the company. She clearly still believes in the thesis that hyperscalers desperately need an alternative to Nvidia to contain costs.
She just looked at the 113% surge, saw the insider selling, and decided, “You know what? The short-term risk-reward ratio is simply better elsewhere.” And this "elsewhere" meant moving from a contender straight to a market leader. The capital withdrawn from AMD was immediately directed into Nvidia on the day the stock fell nearly five percent.
You see, after a period when smaller semiconductor companies and challengers outsold the giants, Wood's maneuver of buying on the downside for $53 million is a huge vote of confidence. This is a powerful signal. She believes that Nvidia is aggressively regaining its absolute dominance in the AI sector, and any significant drop is just a buying opportunity.
And the data perfectly illustrates why Nvidia still wears the crown. Look at this. Of their whopping quarterly revenue of $96.2 billion, a staggering $89 billion came directly from the data center business. The remaining $7.2 billion is the combined income from all other sources. This proves beyond a doubt that the demand for AI is not just a temporary bubble.
It is extremely robust, right down to the level of basic computing. The whole world continues to actively build data centers. What about Nvidia? Well, they are still the ones who supply the vast majority of "engines" . But here's the thing: the Nvidia thesis is no longer limited to hardware. Just days after Arc bought back its falling stock, Nvidia announced the largest acquisition in history, acquiring Hugging Face for $12.93 billion.
If you're not familiar with them , Hugging Face is essentially the main hub for open AI developers. This is a place where the community comes to build and deploy models. As William Blair analyst Sebastian Nagy noted, this is exactly the same as when Microsoft bought GitHub in 2018. This cements Nvidia's huge leap from simply selling chips to fully owning the AI software stack.
They are building an economic moat that will make it incredibly difficult for competitors to challenge them. Along with Nvidia , the second main pillar of this rotation was Broadcom. You can think of them as a true “silent” AI giant. ARK bought over $20 million worth of Broadcom stock right before their historically strong earnings report.
Although Broadcom lagged slightly compared to AMD— up just 7% year-to-date at the close—it was clearly an undervalued asset. Broadcom creates specialized AI accelerators and networking products that are supplied to giants such as Google, Anthropic, and OpenAI. They are essentially the “picks and shovels” in the AI “gold rush,” quietly ensuring the operation of the entire infrastructure that makes this technology possible in the first place .
So what exactly caused this confidence, especially despite the slight drop in stocks due to concerns about the fourth-quarter outlook? It all comes down to management's stunning long-term projections. Just listen. They raised their AI revenue forecast for fiscal year 2026 to $58 billion. They then expect that figure to double to 115 billion in 2027, and double again to a whopping 230 billion in 2028.
This expected annual doubling perfectly explains Wood's incredibly high confidence in Broadcom. This is a bet on basic infrastructure that the broader market has not yet fully appreciated. Okay, now let's move on to the " wild card" of this entire 10-day sprint—the counterintuitive bet. Amidst all these carefully calculated maneuvers in the AI space, ARK invests $44.4 million in Rocket Lab in just two trading days.
We're talking about space infrastructure stocks, which have fallen more than 58% from their all-time highs and are down 42% over the past 90 days. I mean, that's a really bold move, isn't it? But this is a classic Katie Wood move: buying heavily beaten-up, highly volatile stocks on the assumption that they are poised for a massive multi-year rebound.
However, make no mistake. This is a deal with an extremely tight risk-reward profile. On the one hand, you have huge obstacles. Rocket Lab is spending a lot of cash. They just lost a big $ 700 million NASA contract that went to Blue Origin. Their new Neutron rocket is facing production delays. And yes, insiders sell off shares.
Terrible, right? But on the other hand, the arguments in favor of growth are overwhelming. They have a record order backlog of $2.36 billion . They have 94 successful Electron launches, which proves their reliability. And they are making a strategic transition to a vertically integrated space contractor.
Analysts are forecasting an average price target that assumes 70% upside. Wood is clearly looking through the short-term noise to capitalize on the huge long-term opportunity in commercial space . So, when the dust finally settles and we look behind the scenes of these tumultuous 10 days, the masterful strategy will become abundantly clear.
As we said before, this was not a retreat from AI at all. By selling 95 million AMD shares and buying Nvidia, Broadcom, and Cerebras for 124 million, it actually created a net inflow into AI infrastructure worth about $29 million. And at the same time, it strategically diversified its growth investments, adding Block for fintech, heavily backing Rocket Lab for space, and even opening a small position in BWX Technologies for nuclear power.
When we step back and organize all of these steps, the overall turnaround becomes incredibly clear. In the field of AI chips, it went from a contender, AMD, straight to the absolute king, Nvidia. In the field of AI infrastructure, it has moved from software projects like Palantir and gone directly to computing kings like Broadcom and Cerebras.
And in its broader basket of growth investments, it has reduced stakes in life sciences and AI programs to strategically fund these ambitious projects with a high conviction in fintech and space infrastructure. This is truly a master class in reallocating capital in favor of your most confident ideas.
Which leaves us with the main takeaway for your own portfolio. Throughout this explanation, we have seen brilliant aggressive reallocation that favors dominant AI infrastructure providers and deeply undervalued long-term bets. But remember, these are high- conviction trades that can potentially be volatile.
Broadcom faces major customer concentration risks. And Rocket Lab is trading at a price that is almost 50 times its sales volume. So the final question you really have to ask yourself after analyzing all this data is: Are you ready to follow Katie Wood into these particular infrastructure deals, or does this highly concentrated risk profile simply not align with your own long-term strategy? This is certainly worth considering.
Comments 0