Watch stocks with the help of analysts at Golden Qilin: authoritative, professional, timely, and comprehensive—helping you uncover potential theme opportunities!
（Source: NetEase Technology）
The latest filings with the U.S. Securities and Exchange Commission (SEC) show that in the second quarter of this year, the well-known Wall Street growth-oriented hedge fund Tiger Global Management made a major reshuffle of its U.S. stock holdings: it cut back on previously heavily held technology giants and AI popular names, while redirecting capital toward “second-tier” AI chip players, storage chip companies, and payments-related firms, among other sectors.
Tiger Global filed its 13F on Friday, August 14. As of June 30, its total stock holdings had an estimated market value of about $23.982 billion, with positions in 46 individual stocks. Compared with the first quarter, the most notable moves by Tiger Global in the second quarter were initiating positions in two NVIDIA rivals—AI chip maker Cerebras and AMD—while also building stakes in storage chip company Seagate Technology and payments giant Visa, and continuing to add to Intel.
At the same time, Tiger Global also “took profits and locked them in” from the blue-chip tech stocks it previously held heavily. In Q2, all of the top ten holdings—Amazon, Meta, NVIDIA, and Microsoft—were reduced. Among them, Alphabet, the parent company of Google, became the single largest stock by sell-off strength calculated by the value of shares reduced, with a reduced market value of about $1.72 billion. Broadcom, AppLovin, and TSMC followed next.
There were clear changes in the AI chip lineup: Cerebras and AMD became the top two new builds.
Judging by the size of Q2 purchases, Tiger Global’s most aggressive move occurred in the AI chip sector.
Data shows that in Q2 Tiger Global initiated a position in Cerebras of about 3 million shares. Based on the end-of-period market value, it was about $660 million, making it the largest single buy-in for the fund that quarter. At the same time, it initiated a position in AMD of about 675,000 shares, with an end-of-period market value of about $392 million.
Cerebras is a chip company focused on AI accelerators and large-scale computing systems. Its core products compete directly with NVIDIA GPUs. Tiger Global had already invested in Cerebras through the private market before its IPO, and entered the public market further after listing. Shareholder filings disclosed by Cerebras in May this year show that Tiger Global Management holds about 3.496 million shares of Cerebras-related stock.
This means Tiger Global’s Q2 actions were not simply about reducing AI chip exposure; it was more like a “switch of positions” within the AI chip track. On one hand, it reduced holdings in leaders that had already surged a lot, such as NVIDIA and Broadcom. On the other hand, it bet on competitors like Cerebras and AMD that still have room for further growth.
Among them, the new position in AMD is especially worth noting. As one of NVIDIA’s most direct competitors in the AI GPU market, AMD has steadily expanded its data-center AI accelerator business in recent years. Tiger Global established an AMD position of about $392 million in one go, indicating that it still has high confidence in the long-term growth of demand for AI computing power, even though its investment direction has shifted from the absolute market leader to competitors.
At the same time, Tiger Global also increased its holding of Intel by more than 2.6 million shares, worth about $365 million. Unlike Cerebras and AMD, Intel is a major add-on to an existing position; it is also the only stock among the top five increased positions in Q2 that was not a newly initiated position.
In other words, in Q2 Tiger Global adopted a fairly clear “diffusion-style bet” in the chip sector: it bet on changes in the competitive landscape of AI accelerators, and it also bet on the transformation potential of Intel, a traditional CPU giant.
From AI chips to storage: Seagate and Intel became another main thread
In addition to AI computing chips, Tiger Global has also clearly increased its allocation to the AI infrastructure industry chain.
Seagate Technology was a new position in Q2, with an end-of-period market value of about $275 million, ranking fourth on the fund’s buy-in list for the quarter.
Building AI data centers not only boosts demand for GPUs, CPUs, and networking chips, but is also driving demand for high-capacity storage. With the continued expansion of large language model training, inference, and data-center scale, storage and hard disk demand have also become an important component of AI infrastructure investment.
Therefore, from Cerebras and AMD to Intel and then Seagate, Tiger Global’s incremental capital in Q2 was clearly concentrated in AI computing power and its supporting infrastructure industry chain.
Big bets were also made on non-AI assets such as Visa.
Tiger Global did not put all of the newly added funds into tech stocks.
In Q2, the fund initiated a new position in Visa, the payments giant. The market value of the end-of-period holding was about $274 million, ranking fifth on the fund’s list of largest increases for the quarter.
In addition, Tiger Global also increased positions in stocks such as Intuit, Corpay, Reddit, and MercadoLibre.
These actions show that while Tiger Global continues to bet on the AI industry, it is also increasing allocations to growth-oriented assets such as digital payments, financial technology, and internet platforms.
Among them, Reddit and MercadoLibre continue Tiger Global’s long-preferred style of investing in internet platforms, while Intuit and Corpay further broaden its exposure in financial software and payments.
“AI darlings” faced concentrated trimming: Broadcom, AppLovin, and Alphabet were key targets
If the buy side reflects Tiger Global’s bet on a new round of AI opportunities, then the sell side shows that it is cashing out gains on some of the assets that have risen strongly.
In Q2, Alphabet was the single stock with the largest reduction by Tiger Global, with a reduced market value of about $1.72 billion.
Broadcom ranked second, with a reduction of about $690 million. As one of the important beneficiaries during the earlier AI boom, Broadcom won significant attention from the capital markets thanks to its AI data-center networking chips and custom ASIC business. Tiger Global’s major cut to its Broadcom position means it is lowering the concentration of some AI leader companies.
The third-largest reduction was AppLovin. Tiger Global directly exited the position, involving a market value of about $418 million. AppLovin was previously also one of the popular AI concept stocks in the U.S. market, and its advertising technology and AI-driven ability to monetize applications had attracted investors’ interest.
In addition, Tiger Global reduced TSMC by about $327 million and reduced Zillow by about $320 million; the latter ranked as the fifth largest reduction target of the quarter.
TSMC’s reduction is especially worth关注. As a core global company in advanced-process chip manufacturing, TSMC is also an important beneficiary of the AI industry chain. Yet Tiger Global still chose to cut its position, further indicating that its core operations in Q2 were not simply about increasing or decreasing AI exposure, but about reordering assets within the AI industry chain.
Spotify, Netflix, and others were reduced or fully sold off, with holdings further “slimming down.”
Besides the above large-cap tech and AI热门 stocks, Tiger Global also reduced its holding of Spotify.
At the same time, the fund fully exited Netflix, Zscaler, and Procore.
Meanwhile, Netflix’s exit from the holdings means Tiger Global further reduced its allocation to traditional internet platforms and media entertainment assets; Zscaler’s exit means its related positions in the network security field were further trimmed.
Overall, in Q2 the number of positions held by Tiger Global fell from 54 in Q1 to 46, and the total market value of disclosed 13F stock holdings was about $23.982 billion.
All of the top ten holdings were reduced, and the portfolio concentration was adjusted
Based on the market value of end-of-period holdings, the top ten holdings of Tiger Global in Q2 were:
Notably, all of these top ten holdings were reduced by Tiger Global in Q2.
This means that although these companies still occupy core positions in the portfolio, Tiger Global is actively lowering the concentration of the traditional “big tech + AI leader” portfolio, freeing up some capital to invest in companies that previously had lower positions—or had no positions at all.
Looking at the holding structure, this change does not mean Tiger Global is exiting tech stocks. On the contrary, its newly added positions remain highly concentrated in growth tracks such as AI chips, semiconductors, internet platforms, and financial technology.
More precisely, Tiger Global is rebalancing internally in a round of “reducing exposure to leaders while adding to new entrants.”
Comparison of buy and sell lists: from “super leaders” to “winners of the next stage”
If ranked by the market value of positions involved in Q2 trades, the five stocks with the largest reduction amounts in that quarter were:
Rank | Stock | Q2 activity | Involved market value 1 | Alphabet | Significant reduction | $1.72 billion 2 | Broadcom | Significant reduction | $690 million 3 | AppLovin | Fully sold | $418 million 4 | TSMC | Reduced | $327 million 5 | Zillow | Reduced | $320 million
The five stocks with the largest buy-in amounts were:
Rank | Stock | Q2 activity | Involved market value 1 | Cerebras | New position | $660 million 2 | AMD | New position | $392 million 3 | Intel | Increased position | $365 million 4 | Seagate Technology | New position | $275 million 5 | Visa | New position | $274 million
Among them, it is especially worth noting that four of the top five increased-position targets were new positions, while only Intel was an add-on to an existing position.
This structure stands in sharp contrast to the sell side: the sell side is concentrated in Alphabet, Broadcom, TSMC within the NVIDIA supply chain, and AppLovin—assets that have already received high market recognition. The buy side, meanwhile, points more to the next batch of beneficiaries in the AI infrastructure industry chain, such as Cerebras, AMD, Intel, and Seagate.
From this perspective, Tiger Global’s Q2 13F looks more like an “internal rotation map” of AI trading: the fund did not give up on AI. Instead, it reduced positions in some of the super-leaders that had become market consensus, while looking for more resilient chip, storage, and infrastructure companies during the AI capital expenditure cycle.
It should be noted that 13F reflects a fund’s U.S. stock holdings as of June 30, and it is usually disclosed about 45 days after the quarter ends. Therefore, the above data does not necessarily indicate that Tiger Global still holds the same positions as of now, nor can it reflect non-13F assets that were not disclosed, or trading changes since July.