Google issues 80 billion to invest in AI, which companies will benefit?

According to the Zhitong Finance APP, on June 1st, Alphabet (GOOGL.US) dropped a bombshell: it plans to raise $80 billion through equity financing—one of the largest equity financing cases in Silicon Valley history—to supplement ammunition for AI infrastructure expansion. The next day, a "green light" from Omaha also lit up: Berkshire Hathaway, led by Warren Buffett, announced a $10 billion anchor investment, which will be conducted through a targeted issuance at $351.81 per Class A share (approximately a 6% discount). This move undoubtedly increases the heat of AI investment, and analysts have interpreted it.

The true background of the $80 billion equity financing is the super capital expenditure trajectory that Alphabet has already formulated. For the fiscal year 2026, Alphabet expects its total capital expenditure to be in the range of $180 billion to $190 billion, nearly doubling the actual expenditure of $91.4 billion in 2025. Even more astonishingly, according to information previewed by Alphabet's Chief Financial Officer Anat Ashkenazi during the April earnings call, capital expenditure in 2027 will be "significantly higher" than the upper limit of $190 billion for 2026. Bloomberg intelligence analyst Mandeep Singh estimates that Alphabet's capital expenditure in 2027 may surpass the $300 billion threshold, potentially exceeding Alphabet's own operating cash flow scale—injecting $300 billion in physical capital annually to support the AI landscape, an unprecedented move in the history of technology business.

Alphabet is not an isolated case. According to industry data compiled by several institutions, the total AI-related capital expenditure of the four major cloud service providers—Amazon, Alphabet, Microsoft, and Meta—is expected to exceed $700 billion in 2026. Among them, Amazon stands out, with expected capital expenditure reaching $200 billion in 2026; Microsoft's current run rate points to about $150 billion; Meta follows closely, raising its expenditure guidance by 73% to between $115 billion and $135 billion. The influx of such massive capital is leading to the concentration of the three major bottleneck resources: "computing power, electricity, and network resources," which will have an impact throughout the 2030s.

The Supply Chain Frenzy: The "Flow of Money" Logic from TA Semiconductors to Optical Interconnection

The market movements on June 2nd confirmed the release effect of Alphabet's $80 billion capital "valve." TPU supply partner Broadcom (AVGO.US) rose by 4.7%. The two have been deeply bound since the 2010s: Broadcom customizes AI ASICs (Application-Specific Integrated Circuits) for Google and supplies high-throughput Ethernet switch chips for large-scale internal deployment. The 32% single-day surge of Marvell Technology (MRVL.US) is backed by a more critical long-term logic—according to multiple foreign media reports in April, Google has engaged in in-depth negotiations with Marvell Technology to jointly develop new AI memory processing units and some computing modules for the next generation of TPU. If the final agreement is reached, Google will seek a "second blood supply point" for the TPU core architecture apart from Broadcom for the first time, which is a significant incremental opportunity that cannot be ignored for Marvell Technology's new chip business NVIDIA CEO Jensen Huang recently publicly predicted that Marvell Technology could become "the next trillion-dollar company," citing that the demand for optical interconnects and high-speed networks in AI data centers will amplify far beyond market expectations. This infrastructure-level "pipeline" — including Marvell Technology, Coherent (COHR.US), Lumentum (LITE.US), Arista Networks (ANET.US), etc. — will be the biggest beneficiaries during the global computing power expansion cycle due to their leverage effect.

Taiwan Semiconductor Manufacturing Company (TSM.US) also stands to benefit. As the only supplier capable of stably mass-producing advanced process AI chips (from 3nm to the future 2nm and even A16 angstrom-level nodes), TSMC's revenue for the first quarter of 2026 has reached a record $35.6 billion, a year-on-year increase of 35%. It is expected that the compound annual growth rate of AI chip revenue will maintain over 60% before 2029. As Google, Amazon, AMD, and others increase their investments in self-developed chip projects, TSMC has become the "only node" that cannot be bypassed in all competitive roadmaps.

Seeking Alpha analyst YR Research stated: "TSMC remains one of the biggest beneficiaries of Google's growing capital expenditures, and it may be the only company that holds a neutral stance on which model provider or chip manufacturer will win."

Wall Street Perspective

For shareholders, an $80 billion equity financing is a dilution event that cannot be ignored, and Alphabet's stock price plummeting after the announcement intuitively reflects investors' instinctive aversion to EPS dilution. However, more long-term perspectives are attempting to reconcile this conflict. HSBC analyst Paul Rossington maintained a "Buy" rating on Alphabet in a report released on Tuesday, only slightly lowering the target price to $420. He pointed out that Alphabet is in a "unique position to benefit from AI" — no other company can simultaneously possess such a comprehensive advantage in a multi-hundred-billion-dollar advertising cloud base, nearly a decade of accumulated TPU architecture, and a $46.2 billion cloud backlog. According to his estimates, by the end of the first quarter of 2026, Google's cloud backlog (excluding TPU hardware sales revenue) will reach approximately $366 billion, compared to $108 billion in the same period last year — a more than threefold increase that confirms "demand exceeding supply" is not just rhetoric, but reality.

Wells Fargo analyst Ken Gawrelski detailed the flow of cost expenditures. According to his estimates, to meet the surging demand from cloud customers, Google needs to migrate nearly 60% of the computing capacity from its existing fleet for internal use to external customers of Google Cloud. He also predicts that Google's additional computing capacity of 9.3 gigawatts in 2027 may still be "somewhat conservative" and will require further acceleration of installation speed. The tug-of-war between bulls and bears precisely means: the $80 billion is not a depletion of confidence, but a massive clarion call for computing power.

In addition, there is another undercurrent quietly changing the entire capital market's expectations for new stock issuances. This financing by Alphabet is highly likely to initiate a "siphoning effect," diverting funds originally intended for large AI model companies like OpenAI and Anthropic into substantial IPO rounds In institutional investment portfolios, the allocation space for "AI-themed positions" is limited. When Google committed to investing $80 billion in new equity for its self-built computing infrastructure, some fund managers might marginally reduce their exposure to other unlisted or highly valued model vendors—this could become an important potential variable for volatility in the IPO market in the coming months.

Another reaction from Wall Street to Alphabet's financing is the "recalculation of AI economics." Seeking Alpha analyst Julia Ostian pointed out that "even as the AI frenzy recedes and companies cut back on token consumption, cloud service providers can still utilize this computing capacity to serve other customer demands or mitigate their redundancy costs through leasing"—this means that Alphabet's large-scale computing investment is not a win-or-lose gamble, but rather a preemptive investment with an asymmetric risk-reward ratio.

Ostian added, "As for the beneficiaries of this massive capital expenditure, Broadcom and TSMC are undoubtedly the most obvious, with NVIDIA also among them, as it still accounts for a large portion of total capital expenditure. Celestica is another key vendor producing data center hardware for Alphabet. As GPUs and TPUs cannot operate independently, CPUs are beginning to gain attention—AMD and Intel will benefit from this."

However, for shareholders, this highly concentrated cost burden of "preemptive" investments is leading to a profound logical shift in the tech industry. The battle for dominance is accelerating globally from "technological advantage" to "capital advantage." In this war, cash reserves, financing capabilities, and debt capacity may be more important than the specifications of chips themselves. Once the clusters guided by large expenditures are completed, they can still be rented by other vendors; conversely, if massive investments are not made today, there is a high likelihood of facing a computing power shortage and high model latency within three years, leading to a competitive disadvantage. In the critical phase where AI shifts from "who has the largest model parameters" to "who can serve the most users at the largest scale," a single-point technological lead may ultimately be outmatched by the first-mover and scale advantages in infrastructure.

The $80 billion equity financing will not be an isolated case in Silicon Valley. Predictions indicate that by 2027, the combined capital expenditures of the four major tech giants on AI infrastructure are expected to exceed $1 trillion, and could potentially double by the end of the 2020s. For Alphabet, partnering with Berkshire Hathaway to make the most aggressive computing bet in human business history may be a groundbreaking gamble. However, for those suppliers positioned advantageously in the AI computing chain, the capital story of AI is just beginning its prologue