Google Shock | The Full Picture of Anthropic's ¥6.4 Trillion Gemini Dual-Strategy
機械翻訳 / Machine-translated
A
AI Friends
2026年4月27日・54分で読める
機械翻訳 / Machine-translated
@aifriends
AI Friends（https://aifriends.jp）のクロスポスト公式アカウント。AIツールの紹介・使い方・できることを、中学生でもわかるやさしい日本語で届けます。
"Why would Google, which makes Gemini, invest ¥6.4 trillion in rival Claude?" — If that question crossed your mind, you're right to ask.
On April 24, 2026, Google announced an additional investment of up to $40 billion (approx. ¥6.4 trillion) in Anthropic, the maker of Claude.
This article explains — in plain language anyone can understand — the true intent behind the "dual strategy" of simultaneously developing Gemini and pouring massive funds into its biggest rival, and what it means for Japanese businesses.
Let's start by breaking down the news from three angles.
The main players are Google's parent company Alphabet and Anthropic, the maker of Claude.
On April 24, 2026, Bloomberg broke the story, followed by simultaneous coverage from U.S. media including TechCrunch and CNBC.
The total investment is up to $40 billion (approx. ¥6.4 trillion), with $10 billion (approx. ¥1.6 trillion) deployed immediately.
The remaining $30 billion (approx. ¥4.8 trillion) is performance-linked — additional funds invested if Anthropic meets its business targets.
With this investment, Anthropic's valuation has reached $350 billion (approx. ¥56 trillion).
To put it in perspective: Google has placed a ¥56 trillion price tag on a company that exceeds the market capitalization of Toyota, Japan's most valuable company (approx. ¥44 trillion).
And just days earlier, Amazon had announced an additional $25 billion (approx. ¥4 trillion) investment in Anthropic.
An extraordinary situation in which over $65 billion (approx. ¥10.4 trillion) flowed into Anthropic in a single week.
This is the largest deal in the AI industry in 2026.
This investment is a bundled deal of "cash + compute resources."
Google will provide Anthropic with 5 gigawatts of TPU (Tensor Processing Unit) capacity over the next five years.
A "TPU" is Google's proprietary AI chip — a notable alternative to NVIDIA's GPUs.
"5 gigawatts" is equivalent to the power output of five nuclear power plants — a massively large AI compute facility.
Think of it as a compute factory consuming the equivalent of ¥5 billion worth of electricity per month — if a typical household's monthly electricity bill is ¥10,000, then this is 500,000 times that.
Amazon is also providing Anthropic with 5 gigawatts of Trainium (its proprietary chip) capacity via AWS.
Combined, that's 10 gigawatts — Anthropic has completely secured its compute resources from 2026 through 2030.
The risk of "AI development halting due to compute shortages" is now essentially zero.
In the AI industry of 2026, "compute resources" are the scarcest resource of all — and Anthropic, having secured a massive supply, is on rock-solid footing.
We are in an era where compute resources are actually more valuable than cash.
Google's investment is not a standalone news event — it's part of a tectonic shift across the entire industry.
In the same week of April 2026, Amazon also announced an additional investment in Anthropic of up to $25 billion (approx. ¥4 trillion).
Amazon is expanding Claude Platform via AWS, accelerating sales to enterprise customers.
With massive investments from both Google and Amazon, Anthropic has secured a combined total of over $65 billion (approx. ¥10.4 trillion).
It's like a hugely popular idol simultaneously receiving signing bonuses from two major record labels.
Google and Amazon — rivals to each other — have both become Anthropic shareholders and partners, an extraordinary arrangement.
Why are both companies locked in a bidding war? The answer is simple: "losing means taking a massive loss."
As the two-powerhouse structure of OpenAI/Microsoft vs. Claude/Anthropic solidifies, Anthropic — which can play both sides — holds an overwhelming advantage.
There is also strong expectation that Anthropic will IPO within 2026.
A new chapter in the AI infrastructure war has begun.
Here we unpack the biggest mystery — "Why invest in a competitor when you have Gemini?" — from three angles.
Google's strategy is known in AI industry parlance as a "dual-track strategy."
A "dual strategy" means advancing both in-house development and support of external players simultaneously, so that you profit no matter who wins.
Google is developing Gemini (its own model) while also making massive investments in Anthropic (a rival).
It's the same logic as a baseball team owner who strengthens their own players while also buying shares in another competitive team and collecting dividends.
The reason: "The AI industry is a winner-take-most market," and losing means catastrophic losses.
If Gemini wins, Google is king; if Claude wins, Google profits as an Anthropic shareholder.
As of 2026, Claude holds a larger share of the enterprise market than Gemini — so it's rational for Google to "back the winning horse too."
It's the ultimate hedge against downside risk.
And it's a strategic move to go head-to-head against the Microsoft × OpenAI combination.
The second true motive is "lock-in through compute resources."
By providing not just cash but 5 gigawatts of TPU capacity, Google embeds itself deeply into Anthropic's compute infrastructure.
Once Anthropic optimizes for Google TPUs, switching to another cloud provider becomes extremely difficult.
It's similar to a strategy where proprietary battery standards are used in appliances to prevent other manufacturers' products from being used — a classic lock-in play.
Even if Anthropic were to turn against Google in the future, it couldn't run its operations without TPUs.
And every time Claude is delivered through Google Cloud's Vertex AI platform, API usage fees flow back to Google.
Major companies like Coinbase, Cursor, Palo Alto Networks, Replit, and Shopify use Claude via Vertex AI — generating massive daily revenue for Google.
It's a clever revenue model: selling salt to your enemy while charging a salt tax.
A triple-dip of "equity investment," "compute resources," and "API revenue."
A masterpiece of business design that will go down in history.
The third reason is "defending the advertising business in the AI era."
About 80% of Google's revenue comes from search advertising — if that is eroded by AI (ChatGPT, Perplexity, etc.), its entire business model collapses.
As of 2026, more people are using ChatGPT and Claude for research, and Google Search usage is gradually declining.
It's the same structural shift as department store sales being slowly eaten away by e-commerce.
While Google fights back with Gemini, it is converting its position to one where it "profits from the AI market as a whole" via its Anthropic investment.
Evolving from "monopoly on search advertising" to "one of the key players in the AI market."
Concretely, the plan is to offset declining search ad revenue by growing Google Cloud sales.
In Q1 2026, Google Cloud revenue grew 38% year-over-year, driven by AI demand.
The Anthropic investment also makes sense as a way to secure Google Cloud's largest customer.
A pivotal transformation from "ad company" to "AI infrastructure company."
It mirrors Microsoft's Azure × OpenAI strategy exactly — achieved by pulling Claude away from Azure.
This is the moment the competitive landscape changes decisively.
Let's map out the 2026 AI industry power structure across three frameworks.
The industry now has two major camps: "Microsoft × OpenAI" and "Google × Anthropic."
Microsoft × OpenAI is an "exclusive partnership" — Microsoft was the sole provider of OpenAI via Azure (until early 2026).
Google × Anthropic, by contrast, is a "non-exclusive partnership" — Anthropic has contracts with both Amazon and Google.
In relationship terms, Microsoft × OpenAI is like a marriage, while Google × Anthropic is like an open, multi-partner arrangement.
Anthropic's strength is its "independence" — not tied to any single company, serving multiple customers.
In 2026, Microsoft announced the end of exclusivity, opening OpenAI to other clouds including Google Cloud.
As a result, the gap between OpenAI and Anthropic is narrowing, and competition is intensifying.
The dynamic is shifting from a battle between two camps to a matrix of two models (OpenAI/Anthropic) × two clouds (Azure/Google Cloud).
For customers, this means more options — a welcome development.
2026 is the year that "AI model × cloud pairing flexibility" accelerates.
Another major player is Amazon, operating through its cloud giant AWS.
In April 2026, Amazon announced an additional investment of up to $25 billion (approx. ¥4 trillion) in Anthropic.
Amazon's cumulative Anthropic investment now totals approximately $15 billion (approx. ¥2.4 trillion).
Amazon is supporting Anthropic's training via its proprietary Trainium chips through AWS, while deploying Claude Platform at scale.
It's a battle for dominance — if Google takes Tokyo Station, Amazon is waiting at Shinjuku.
The result: Anthropic, receiving funding and compute from both clouds, is now in an invincible position.
Anthropic CEO Dario Amodei commented: "Risk is diversified through a multi-cloud strategy."
Within 2026, Anthropic will be available on all major clouds — AWS, Google Cloud, and Azure.
The era where "Claude can be used on any cloud" has arrived.
For enterprise customers, this means broader options and accelerating price competition — a favorable environment.
Not all Big Tech is moving in the same direction — companies pursuing independent paths remain very much alive.
Meta is releasing "Llama" as open source, integrating it into its own services (Instagram, WhatsApp).
Elon Musk's xAI is independently developing "Grok," linked to X (formerly Twitter).
In April 2026, xAI announced SpaceX GPU in-house production and plans for a ¥260 trillion IPO, raising massive capital on its independent path.
It's like Sony and Panasonic partnering in consumer electronics while a garage startup competes with its own independent development.
The result: the 2026 AI industry has a three-polar structure — "two major camps + independent players."
Anthropic (Google/Amazon camp), OpenAI (Microsoft camp), Meta/xAI (independent path).
China's DeepSeek also released "V4 Flash/Pro," going on the offensive with open models.
Multi-polarization intensifies competition, with companies vying on technology and price.
For users, a wealth of options and falling prices — a favorable environment.
The power map is expected to shift further in the second half of 2026 through 2027.
Why is such massive investment flowing in? Let's explain Anthropic's capabilities with three data points.
On April 7, 2026, Anthropic announced that its "Annual Revenue Run Rate (ARR) has exceeded $30 billion (approx. ¥4.8 trillion)."
"ARR" stands for "Annualized Run Rate" — monthly revenue multiplied by 12 to estimate annual revenue.
At the end of 2025, Anthropic's ARR was approximately $9 billion; in just four months, it more than tripled.
OpenAI's ARR during the same period was approximately $25 billion — Anthropic has overtaken it in revenue for the first time.
It's like the second-place team suddenly overtaking first place, with momentum to extend the lead in the remaining matches.
The growth was driven by "Claude Code," an AI-powered code generation tool.
Developers subscribing to Claude Code for hundreds of dollars a month; programmers worldwide have embraced it.
Large enterprise contracts are also surging — the number of contracts worth $1 million or more annually doubled from 500 to over 1,000 companies.
A "developer × enterprise" dual-track strategy paying off.
An ARR exceeding $50 billion within 2026 is within sight.
The moment the AI industry's power map changed completely.
With this Google investment, Anthropic's valuation has reached $350 billion (approx. ¥56 trillion).
At the time of the Series G round in February 2026, the valuation was $380 billion — maintained into the April investment.
However, Bloomberg reports that "offers from investors exceeding $800 billion are pouring in."
Anthropic's CFO has expressed caution: "We do not plan additional fundraising at this time."
It's like a situation where you're not selling your house, yet local real estate agents are swarming with offers at double the asking price.
The reasons: the "pace of growth" and "IPO expectations for 2026–2027."
If it IPOs, there's a possibility of a market capitalization exceeding $1 trillion (approx. ¥160 trillion).
OpenAI's valuation as of October 2026 is $500 billion, closing in on Anthropic.
Both companies leading the industry as the world's largest AI firms.
Google and Amazon, who can become shareholders before the IPO, can expect massive unrealized gains.
The latter half of 2026 through 2027 has the feel of an "AI company IPO rush."
The biggest driver of rapid growth is "Claude Code," an AI-powered code generation tool.
"Claude Code" is a terminal-native coding AI developed by Anthropic.
Rapidly gaining market share in the developer space, surpassing rivals like GitHub Copilot and Cursor.
As of April 2026, monthly developer subscription fees are the primary revenue source, used by tens of millions worldwide.
It's the most coveted ticket in the programmer world — like the hottest stadium entry pass.
The reasons: "code quality" and "ability to understand large-scale projects" — the Opus 4.7 model is particularly dominant.
Claude Opus 4.7 is now generally available via Vertex AI, usable securely by Google Cloud users.
In the second half of 2026, Claude Opus 5 is expected — further performance improvements ahead.
Anthropic's strategy is a bottom-up approach: "secure developers first, then secure enterprises."
This strategy has worked beautifully, delivering ARR exceeding $30 billion.
Going forward, enterprise contracts will be the growth engine.
2026 is the turning point in the "developer AI war."
Let's consider the relevance to Japanese business from three angles.
The biggest piece of good news for Japanese companies is that "Claude Opus 4.7 is available via Google Cloud Vertex AI."
"Vertex AI" is Google Cloud's AI platform, which also offers Anthropic's Claude.
In April 2026, Claude Opus 4.7 became generally available on Vertex AI.
New features include "support for high-resolution image input" and "improved accuracy for complex documents and graphs."
The feeling: the latest AI assistant is now available in a trusted Google Cloud package.
Japanese companies with existing Google Cloud contracts can use Claude without any additional contract.
Data can be processed in Japanese regions (Tokyo, Osaka) — no worries about security or compliance.
NEC, Toyota, and SoftBank are already exploring the Google Cloud × Claude combination.
"Freedom to choose AI models" has expanded dramatically at once.
2026 could be the inaugural year for Claude adoption among Japanese enterprises.
The fundamental benefit of this investment is "an increase in options."
Through 2025, the AI choice for Japanese companies was dominated by "OpenAI (via Azure)" — everything else was limited.
From 2026 onward, free selection from combinations of "OpenAI (Azure/Google Cloud)," "Claude (AWS/Google Cloud)," and "Gemini (Google Cloud)" becomes possible.
Optimal AI can be chosen for each purpose, with costs optimized through comparison.
The ease of choosing an AI model feels like picking a bento at a convenience store.
Example: Use Claude for coding, Gemini for image generation, GPT-5.5 for general chat — each tool for its best use case.
Many large Japanese enterprises are beginning to adopt a "multi-model strategy."
Major SIers like Fujitsu, NTT Data, and Hitachi are also offering multi-cloud, multi-model AI construction services.
Intensifying competition is driving prices down — API usage fees have fallen 30–50% compared to 2025 in some segments.
Even small and medium-sized businesses can leverage AI at lower cost.
2026 is the inaugural year of "AI democratization."
Another impact: "activation of the domestic infrastructure and talent market."
Anthropic plans to invest $500 billion (approx. ¥8 trillion) in domestic data centers within 2026.
A portion may be invested in Asia including Japan, in tandem with expansion of Google Cloud's Tokyo and Osaka regions.
Data center construction drives increased demand for power and cooling equipment — a tailwind for infrastructure companies like Tokyo Electric Power and Kansai Electric Power.
The ripple effect: like a new industrial park being built, benefiting surrounding construction companies, utilities, and logistics firms.
At the same time, demand for AI talent is exploding — Anthropic, Google, and AWS are ramping up engineer recruitment in Japan.
As of April 2026, the median annual salary for AI engineers in Tokyo is ¥15 million, with top earners exceeding ¥50 million.
The mid-level engineer job market is also booming — those with "AI × Japanese language" skills are in extremely high demand.
The Anthropic × NEC talent development partnership (announced April 2026) is also accelerating domestic AI talent cultivation.
IT, energy, education, and talent sectors — economic ripple effects in all directions.
A turning point at which Japan's AI industry begins to rise in earnest.
Kenta is a CTO at a Tokyo startup who was struggling at the stage of embedding AI features into a SaaS product.
His situation: "I built everything with OpenAI, but I want something smarter for coding-related features."
When he saw the news about Google × Anthropic's ¥6.4 trillion investment, he decided to try Claude via Vertex AI.
No additional cost on top of the existing Google Cloud contract — he connected Claude Opus 4.7 via API.
Result: The accuracy of coding support features improved by 30% compared to OpenAI, and user satisfaction soared.
The feeling: "Like a chef who started using five different knives for different purposes."
And the pricing is pay-as-you-go based on API usage — adding Claude costs roughly ¥200,000 per month.
A "multi-model design" uses the optimal AI for each function, boosting technological competitiveness.
Differentiation from competitors (startups using only OpenAI).
A few months later, in a funding round, the "multi-model AI strategy" was recognized — he successfully raised ¥2 billion.
Embodying the new standard for startups in 2026.
Kana handles AI promotion at a manufacturing company in Osaka and was struggling with AI governance across the organization.
Her challenge: "Different departments use different AI services, and I can't manage data leakage risks."
The news prompted her to decide to consolidate onto Google Cloud Vertex AI.
With Vertex AI, Gemini, Claude, and (some) OpenAI models can all be managed on a single platform.
The effect: like unifying separate mobile phone contracts by department into a single-company family plan.
Result: AI usage across departments became visible, and costs were reduced by 20% through unified management.
The burden on the IT department dropped significantly, with usage requests and audits centralized.
A dedicated "AI Governance Committee" was established, improving quality through monthly usage reviews.
Six months later, in-house AI usage tripled, and the business efficiency gains were recognized by senior management.
"Multi-model × unified management" is the new gold standard for large enterprise AI adoption.
The standard approach for large enterprise DX managers in 2026.
Takuya is a freelance engineer in Fukuoka who was working on an independent SaaS app development project.
His situation: "Claude Code doubled my development efficiency, but the $200/month personal plan is a burden."
When the Google × Anthropic investment news came out, he started using Claude via Vertex AI.
With pay-as-you-go API pricing, keeping monthly usage down brings the bill to around $50/month.
And using Google Cloud's free tier for personal projects makes the first month essentially free.
The feeling: "Like using up all the free coupons handed out on the street to stretch your living budget."
Cost optimization gives him the breathing room to develop multiple projects in parallel.
Two months later, he launched his SaaS app — it went viral on X (formerly Twitter), and he hit ¥500,000 in monthly revenue in his first month.
Even individual developers can turn "freedom to choose AI models" into a powerful weapon.
"Strategies used by large corporations" also benefit individuals — a real sense of AI infrastructure democratization.
The moment the barrier to AI adoption dropped for freelancers and side hustlers.
2026 heralds the golden age for individual creators.
A. It's a "dual strategy" aimed at "profiting no matter who wins."
The AI industry has a strong tendency toward "winner-take-most," and losing means catastrophic losses.
If Gemini (its own product) wins, Google is king; if Claude (Anthropic) wins, Google profits as a shareholder.
It's the same logic as a baseball team owner who cheers for their own team while also buying shares in other clubs and collecting dividends.
Moreover, the investment is a bundled deal of "cash + compute resources" — by providing TPU capacity, Google also grows Google Cloud revenue.
Additionally, API usage fees for Claude via Vertex AI flow back to Google — a triple-dip.
The result: a revenue structure that doesn't depend on "who wins among the AI models."
The strongest possible risk hedge against uncertainty in the 2026 AI industry.
A flexible, non-exclusive strategy in contrast to the Microsoft × OpenAI exclusive model.
A clever design of "selling salt to your enemy while charging a salt tax."
A decision that will go down in business history.
A. It's a scale that would rank in the top tier of Japanese corporate revenues — and this is for a single company investment.
¥6.4 trillion is more than double the 2026 net profit of SoftBank Group (approx. ¥3 trillion).
It's equivalent to about 1% of Japan's GDP, or about 80% of Tokyo Metropolitan Government's annual budget (approx. ¥8 trillion).
To put it in perspective: "Google invested the equivalent of 80% of Tokyo's annual spending into a single startup."
And combined with Amazon's additional investment (approx. ¥4 trillion), Anthropic raised approximately ¥10.4 trillion in a single week.
This ranks among the world's largest M&A and investment deals of 2026, comparable to major semiconductor transactions.
Historically, it is among the largest-ever investments in a single startup.
Second only to Microsoft's investment in OpenAI (equivalent to approx. ¥13 trillion).
A figure that symbolizes the extraordinary concentration of capital in the AI industry.
2026 will be remembered in history as "the year massive money flowed into AI."
The fact that an investment of this scale was executed itself demonstrates the enormous expectations for the AI market.
A. The main options are "via Vertex AI" or "via Amazon Bedrock" — choose based on your situation.
Companies already on Google Cloud: Vertex AI is the fastest route, no additional contract required.
Companies already on AWS: Amazon Bedrock, also a secure environment.
Both clouds support Japanese region (Tokyo, Osaka) processing — no data cross-border risk.
It's as simple as choosing between FamilyMart and 7-Eleven — whichever is closer works fine.
How to get started concretely: ① Log in to Google Cloud Console (or AWS Console), ② Enable Vertex AI (or Bedrock), ③ Access the Claude model and connect via API.
Japanese-language technical documentation is well-established, and technical articles on G-gen blog, Zenn, and Qiita are plentiful.
Starting with prototype development using the free tier is recommended.
API usage is pay-as-you-go — small starts are possible from a few thousand yen to tens of thousands of yen per month.
The key is consulting with your in-house AI team and gradually expanding usage.
In 2026, simply "trying it out" offers substantial first-mover advantages.
A. There are currently no acquisition plans, and Anthropic intends to maintain its independence.
This investment is an "equity stake," not an acquisition — Google is a minority shareholder.
Anthropic CEO Dario Amodei has publicly stated that "independence is the top priority."
The company has been registered as a PBC (Public Benefit Corporation) since 2025, pursuing social objectives alongside shareholder value.
It's similar to a major publisher providing funding to an up-and-coming author while preserving the author's creative freedom.
Furthermore, by also receiving massive investment from Amazon, Anthropic maintains a balance that prevents it from being tied to any single company.
There is strong expectation that Anthropic will IPO (go public) in 2026–2027, at which point it will be a fully independent listed company.
After the IPO, Google and Amazon will remain major shareholders with influence, but acquisition would be difficult from an AI regulatory perspective (U.S., EU, Japan).
The relationship is "strategic partner," not "subsidiary."
This independence is Anthropic's strength — the reason it continues to attract top researchers.
Unlike OpenAI, there is no turbulence from a "nonprofit → for-profit conversion," and management is stable.
The most watched independent player in the 2026 AI industry.
A. Yes — in fact, individuals and small businesses stand to gain the most.
Pay-as-you-go API pricing means serious AI adoption is possible from just a few thousand yen per month.
Concretely: Claude Code (code generation) from $20/month, Claude.ai (chat) from $20/month.
Via Vertex AI, there's also a free tier — you can start for just a few hundred yen per month.
It's the democratization of fine dining, now available casually as a deli takeaway.
Concrete use cases: building a website for a sole proprietor, an inquiry response AI for a small business, meeting minute summarization for freelancers.
The mainstream approach is not "hiring an AI engineer" but "subscribing to an API and using it with prompts."
Technical barriers are also lowering — subscribing to a service for a few tens of thousands of yen per month enables adoption without specialized knowledge.
Price competition among competitors (OpenAI, Google, Anthropic) has driven API usage fees down 30–50% compared to 2025.
For individuals and small businesses that want to "improve work efficiency with AI," 2026 is the best possible timing.
Start by trying the free plan to see if it fits your business — then expand from there.
First-mover advantages are still very much available.
In April 2026, Google's "dual strategy" of investing ¥6.4 trillion in rival Anthropic rewrote the common sense of the AI industry in one stroke.
Developing Gemini while also investing in Claude — a clever design that pays off no matter who wins.
And with a 5-gigawatt TPU contract, Google Cloud also captures revenue on the compute resource side.
"Selling salt to your enemy while charging a salt tax" — a judgment that will go down in business history.
An event that embodies "a new-era game theory where rivals join hands."
For Japanese companies too, Claude is now accessible via Vertex AI and Amazon Bedrock, dramatically expanding freedom of choice in AI models.
An era where startups, large enterprises, and individual developers alike can all benefit from a multi-model strategy.
Your project today might open up new possibilities simply by logging into Google Cloud (or AWS) and trying Claude — don't miss the wave of the AI infrastructure revolution; take that first small step now.
This article is a cross-post from AI Friends.
