Anthropic's $30 Billion Revenue, Samsung's 755% Profit Surge—The "Now" of the AI Semiconductor Value Chain and Tips for Investment Decisions
In April 2026, news surrounding the AI semiconductor value chain broke all at once.Anthropic expanded its Google TPU contract to the gigawatt scale, reaching an annual revenue run rate of $30 billion.Samsung reported a quarterly profit increase of 755% year-over-year, setting a new record. Meanwhile, in chip procurement, diversification away from Nvidia's dominance is accelerating across all players. Based on Bloomberg Tech's reporting, we summarize the front lines of the AI economy and the key points investors should grasp.
1. Anthropic—$30 Billion Annual Revenue and Massive TPU Expansion
1-1. Claude Code and Co-Worker Drive Demand
According to Bloomberg Tech reporter Shirin Ghaffary, Anthropic has expanded its TPU contract with Google to the gigawatt scale. The contract, which was initially 1 gigawatt, has grown more than threefold.
"The popularity of Claude Code (coding agent) and Claude Co-Worker (general-purpose agent) is skyrocketing, and enterprise growth is extremely strong. Expanding computing resources has become an urgent priority."
Anthropic's annual revenue run rate has reached $30 billion. This scale is boosting the entire ecosystem of the TPU and custom chip business, including Broadcom.
1-2. Growth Continues Despite Dispute with the Pentagon
Anthropic is currently involved in a lawsuit with the U.S. Department of Defense. In its complaint, the company claims that this dispute could lead to "loss of customers and billions of dollars in revenue risk." However, reporter Ghaffary noted that "at least for now, it has not stopped revenue growth." The competitiveness of the product itself is creating demand that outweighs geopolitical risks.
1-3. Adversarial Model Distillation—Unusual Cooperation Among Three U.S. Frontier Labs
Anthropic, OpenAI, and Google—three companies that usually compete fiercely—are engaging in rare cooperation on the challenge of Adversarial Model Distillation.
"The technique of using the output of old models to train new models is done routinely. The problem is that companies in countries like China are copying U.S. companies' models and releasing them as free open models."
The three companies are sharing information on how to detect this practice. This move indicates that the front line of the U.S.-China AI race is expanding beyond model performance to include the defense of intellectual property.
2. Samsung—755% Quarterly Profit Increase, Approximately 90% from Memory
2-1. Preliminary Figures Exceed Expectations
Samsung recorded a record-high operating profit in its preliminary figures for the January-March 2026 quarter, an approximately eightfold increase (755% growth) year-over-year. It exceeded even the market's high expectations, effectively surpassing the entire 2025 annual profit in just one quarter.
According to analyst estimates, approximately 90% of the 38 trillion won in operating profit comes from the memory chip division. This is the result of explosive growth in demand for High Bandwidth Memory (HBM) for AI data centers.
2-2. A Sudden Turnaround from a Negative Narrative
As Bloomberg Tech reporters pointed out, until very recently, the dominant negative view was that Samsung was "lagging significantly behind SK hynix in HBM."
"It is interesting how quickly they turned the narrative around"
This sudden turnaround confirms that the breadth of AI semiconductor demand is providing structural benefits not only to TSMC and Nvidia but also to memory manufacturers. Note that as these are preliminary figures, details by division are expected to be revealed in the full earnings report on April 30.
3. Multipolarization of the Chip Supply Chain—The Big Picture of "De-Nvidia"
3-1. All Players Accelerating Diversification
Looking across Bloomberg Tech reports, the diversification of chip procurement by hyperscalers and frontier labs emerges as a clear trend.
Gil Lurie of D.A. Davidson summarizes this composition.
"All of the hyperscalers and frontier labs are diversifying into the major chipmakers, trying not to be too dependent on Nvidia. They don't even know what the mix is going to be yet. The chip companies that have the most execution will get the volume."
3-2. Broadcom's Benefit
One of the biggest beneficiaries of this multipolarization is Broadcom. In addition to the expansion of custom chip supply for Google TPUs, Anthropic's rapid growth is directly leading to increased orders. At the time of the Bloomberg Tech broadcast, Broadcom stock was up about 4%.
4. ARM—A "Multi-Order" Shift from Smartphones to Data Centers
The CEO of ARM discussed the dramatic change in their business composition in an interview with Bloomberg Tech Euro.
"In a few years, cloud and AI data centers will be our largest business, and in five years, it will be orders of magnitude larger than smartphones."
ARM is aiming to enter the $10 billion data center market, but on the other hand, Morgan Stanley downgraded the company to equal-weight (neutral), expressing concerns about execution risk. The shift from a stable revenue source like smartphone licensing to a completely different business model of selling data center chips offers significant upside if successful, but the road is not smooth.
5. Macro Environment and Investment Strategy—"Don't Chase, But Be Prepared"
5-1. Middle East Risk Pressures Tech Stocks
On the day of the Bloomberg Tech broadcast, the market reacted strongly to tensions between the U.S. and Iran. The Nasdaq 100 fell 1.5%, and oil prices rose. Reports that Iran had broken off negotiations ahead of the 8:00 PM deadline set by President Trump amplified anxiety.
5-2. Views of Janet Mui of Brewin Dolphin
Janet Mui of asset management firm Brewin Dolphin made her cautious stance on tech stocks clear.
"If oil prices remain high, it will bring growth pressure to the global economy and upside risks to inflation. The central bank's rate-cut agenda will be abandoned, and even rate hikes are possible. You shouldn't chase last week's rally."
She recommends three measures for portfolio defense.
Energy stocks: Inflation hedge + high dividends
Gold: It has fallen from its recent highs and is a relatively good entry point
TIPS (Treasury Inflation-Protected Securities): A direct hedge against high inflation outlooks
5-3. However, continue monitoring AI fundamentals
On the other hand, Mr. Mui also stated the following.
"The themes are progressing in parallel. Some high-quality growth stocks and AI infrastructure 'picks and shovels' stocks have seen significant discounts in recent months. If volatility leads to further discounts, it will be an opportunity to add to positions."
Tech stock exposure remains centered on the U.S., and he maintained the view that 'the foundational layer of AI infrastructure (picks and shovels) is a structural growth driver, regardless of software obsolescence.'
6. VC Frontline—Re-industrialization and 'Non-obvious AI Utilization'
6-1. Eclipse: $1.3 billion for physical industries
VC firm Eclipse has raised $1.3 billion across two funds for startups in the robotics, manufacturing, and energy sectors. Founder Lior Susan spoke from Washington D.C.
"Physical industries are essential to raising the U.S. GDP growth rate to 5%. Physical industries account for 85% of global GDP. The convergence of five forces—capital, policy, technology, talent, and customer demand—has not been seen in U.S. history since the era of Henry Ford and Carnegie. Previously, only China had all of these in place."
Mr. Susan is scheduled to appear with J.P. Morgan's Jamie Dimon to discuss the Defense Production Act, stating, 'J.P. Morgan is committing $1.5 trillion to support these businesses.'
6-2. FPV Ventures: Investing in 'AI companies that don't look like AI companies'
Wesley Chan of FPV Ventures (founding partner of Google Ventures) describes his investment philosophy as follows.
"Not all AI winners look like AI companies from the start"
The symbol of this is Canva.
"Canva is the third most common way for people under 35 to access AI. ChatGPT is first, Google is second. Most people think of Canva as a design company, but it is now an AI company."
Mr. Chan is also investing in the AI drug discovery sector, and one of his portfolio companies has discovered a drug that uses AI to regenerate knee cartilage.
"AI predicted the results of clinical trials and found the drug. It has the potential to make knee replacement surgery unnecessary for skiers and runners."
His investments are spread across the U.S. (Boulder, Boston, New York, etc.), and he said, 'As long as they are trying to change the world in a big way, the location doesn't matter.'
