Amazon Nears $7.5 Billion Acquisition of Israeli AI Infrastructure Startup Decart
Amazon is in advanced talks to acquire AI efficiency firm Decart for $7.5 billion, aiming to reduce its reliance on Nvidia hardware. The deal would mark one of the tech giant's largest strategic investments in custom silicon and model optimization.
Share this story
Amazon is in the final stages of negotiating a $7.5 billion acquisition of Israeli artificial intelligence startup Decart. The agreement, expected to be sealed within days, represents a massive push by the e-commerce and cloud computing giant to control its own AI infrastructure.[1][2]
The proposed transaction would mark one of Amazon's largest strategic purchases to date, aimed specifically at boosting AI chip efficiency across its massive server network. By integrating Decart's technology, Amazon Web Services hopes to significantly reduce the computational overhead required to train and run large language models.[4][5]
"This is a definitive move by Amazon to own the efficiency layer of generative AI," notes the analysis from explainx.ai, which verified the core claims of the ongoing negotiations on November 21, 2026. The $7.5 billion valuation underscores the immense premium cloud providers are currently placing on hardware optimization.[7]
The deal arrives during a critical bottleneck in the global technology sector, where the sheer cost of processing power has constrained AI development. Decart has built its reputation on software and hardware co-design that squeezes maximum performance out of existing silicon architectures.[3][4]
Bypassing the Nvidia Bottleneck
Amazon's aggressive move is widely interpreted as a direct challenge to Nvidia, the undisputed leader in artificial intelligence hardware. Currently, Amazon Web Services relies heavily on Nvidia's expensive graphics processing units to power its enterprise cloud offerings, a dependency that cuts deeply into profit margins.[4]
By bringing Decart's optimization technology in-house, Amazon aims to make its own custom silicon—such as its Trainium and Inferentia chips—more competitive. This vertical integration could eventually allow Amazon to offer cheaper AI processing rates to its cloud customers than rivals relying strictly on external hardware.[4][5]
Interestingly, the path to this acquisition was not uncontested, though other major players ultimately stepped back. According to Dealroom, both Nvidia and the prominent AI research firm Anthropic previously evaluated Decart but chose to pass on an acquisition earlier in 2026.[6]
Anthropic's decision to walk away is particularly notable given the startup's deep existing ties to Amazon, which holds a massive multi-billion dollar stake in the AI developer. Nvidia's pass suggests the chipmaker may not have viewed Decart's specific optimization approach as complementary to its own tightly integrated CUDA ecosystem.[6]
For Amazon, however, the calculus is entirely different. The company operates at a scale where even single-digit percentage improvements in computational efficiency translate to billions of dollars in saved energy and hardware costs over a multi-year horizon.[1][4]
A Milestone for Israeli Innovation
If finalized, the $7.5 billion price tag will make the Decart acquisition one of the most lucrative exits in the history of the Israeli technology sector. The startup emerged from a dense ecosystem of military and academic engineering talent focused on deep hardware optimization.[2][6]
CTech reports that the two companies are set to seal the deal "within days," capping off a rapid negotiation period that accelerated throughout November 2026. The swift timeline reflects Amazon's urgency to secure the talent and intellectual property before competing cloud providers could mount a counteroffer.[2]
Decart's engineering teams are expected to be integrated directly into the Amazon Web Services hardware division. This unit has been under immense pressure to deliver next-generation infrastructure capable of supporting the explosive enterprise demand for generative artificial intelligence.[4][5]
The acquisition also highlights a broader trend of hyperscale cloud providers looking beyond Silicon Valley for critical infrastructure innovations. Tel Aviv and Haifa have increasingly become focal points for semiconductor and systems-level engineering talent, drawing billions in foreign direct investment.[2]
Shifting the Balance of Power
Financial markets have been closely monitoring the negotiations, as the sheer size of the deal signals Amazon's willingness to deploy its massive cash reserves to win the AI infrastructure war. Seeking Alpha noted that the $7.5 billion valuation sets a new benchmark for startups operating purely in the optimization layer.[3]
The transaction could trigger a wave of defensive acquisitions by Microsoft and Google, both of which are similarly racing to reduce their dependence on third-party silicon. As the cost of training frontier models approaches the billions, efficiency has become the primary battleground for cloud supremacy.[1][4]
"Amazon is essentially buying time and talent to accelerate its custom silicon roadmap," the Financial Times reported, highlighting how the deal fits into the broader strategy of the Seattle-based technology giant. The integration of Decart's technology could shave months or even years off Amazon's hardware development cycles.[1]
Regulatory scrutiny remains a potential hurdle, given the intense focus antitrust authorities in the United States and Europe have placed on artificial intelligence consolidation. However, because Decart is an infrastructure optimization firm rather than a consumer-facing AI model developer, the path to approval may be smoother.[7]
The Technical and Financial Stakes
The technical foundation of Decart’s appeal lies in its novel approach to memory bandwidth management. By restructuring how data flows between processors during the training phase, the startup has demonstrated the ability to cut energy consumption by nearly 30 percent on standard workloads.[4][7]
That 30 percent reduction is a massive operational advantage for Amazon Web Services, which currently operates dozens of hyperscale data centers globally. Electricity costs and thermal management have become the primary limiting factors in expanding artificial intelligence compute capacity over the past two years.[3][4]
Amazon’s aggressive capital expenditure strategy was previewed during its third-quarter earnings call in October 2026, where executives signaled a willingness to spend heavily on infrastructure. The $7.5 billion Decart purchase aligns perfectly with the company’s stated goal of prioritizing long-term technological dominance over short-term margin expansion.[1][3]
Microsoft Azure and Google Cloud will now face increased pressure to demonstrate their own hardware optimization strategies to enterprise clients. If Amazon can successfully leverage Decart’s technology to lower the cost of AI inference, it could force a broader price war across the cloud computing sector.[4]
Integrating the Technology
The immediate challenge for Amazon will be successfully merging Decart's highly specialized engineering culture with its own massive corporate structure. Retaining the key architects who built the startup's core technology will be crucial to realizing the $7.5 billion investment's value.[2][5]
Industry analysts expect Amazon to begin rolling out Decart-optimized cloud instances to its enterprise customers by late 2027. These specialized servers will likely be marketed as a cost-effective alternative to standard GPU clusters for specific machine learning workloads.[4]
The deal also raises questions about the future of independent infrastructure startups. With Amazon setting such a high acquisition floor, venture capitalists may increasingly fund hardware optimization companies with the explicit goal of selling them to the major cloud hyperscalers.[3][6]
The finalization of the deal will require sign-offs from multiple regulatory bodies, including the Federal Trade Commission in the United States. However, legal experts suggest the vertical nature of the acquisition makes it less vulnerable to the antitrust challenges that have derailed horizontal mergers in the tech space.[7]
The finalization of the deal will require sign-offs from multiple regulatory bodies, including the Federal Trade Commission in the United States.
As the November 2026 negotiations enter their final hours, the technology industry is watching closely to see how the integration will be structured. The outcome will provide a clear blueprint for how the next generation of artificial intelligence infrastructure will be built, funded, and ultimately consolidated.[1][2]
Key points
- Amazon is finalizing a $7.5 billion agreement to purchase Decart, an Israeli startup specializing in AI compute efficiency.
- The acquisition aims to bolster Amazon Web Services' custom silicon capabilities and directly challenge Nvidia's market dominance.
- Both Anthropic and Nvidia previously evaluated and passed on acquiring the startup before Amazon advanced its bid.
- The deal is expected to close within days, marking a significant consolidation in the AI infrastructure sector.
What we don’t know
- Whether the Federal Trade Commission or European regulators will attempt to block the acquisition on antitrust grounds.
- The specific timeline for when Decart's optimization technology will be available to AWS enterprise customers.
- How much of the $7.5 billion valuation is structured as cash versus Amazon stock.
How we got here
Early 2026
Anthropic and Nvidia evaluate Decart for potential acquisition but ultimately pass.
October 2026
Amazon signals aggressive infrastructure spending during its third-quarter earnings call.
Mid-November 2026
Amazon enters advanced negotiations to acquire Decart for $7.5 billion.
November 22, 2026
Reports indicate the deal is set to be sealed within days.
- Cloud Hyperscalers
- View custom silicon and deep hardware optimization as the only sustainable path to scaling AI infrastructure profitably.
- Incumbent Chipmakers
- Argue that tightly integrated, proprietary hardware ecosystems remain the most powerful and reliable solution for frontier AI training.
- Startup Ecosystem
- See the massive valuation as validation that infrastructure optimization is the most lucrative sector for early-stage investment.
- Industry Analysts
- Focus on the regulatory and integration challenges of absorbing highly specialized engineering teams into massive corporate structures.
Perspectives this story doesn't cover
- Enterprise cloud customers
- Antitrust regulators
Sources
[1]Financial TimesCloud HyperscalersAmazon is nearing a $7.5bn deal to buy start-up Decart
Read on Financial Times →
[2]CTech by CalcalistStartup EcosystemAmazon set to seal Decart deal worth up to $7.5 billion within days
Read on CTech by Calcalist →
[3]Seeking AlphaStartup EcosystemAmazon nears $7.5B deal to buy AI start-up Decart
Read on Seeking Alpha →
[4]LivemintCloud HyperscalersAmazon eyes $7.5 billion Decart deal to boost AI chip efficiency, challenge Nvidia
Read on Livemint →
[5]The Chosun DailyCloud HyperscalersAmazon to Acquire AI Startup Decart for $7.5 Billion
Read on The Chosun Daily →
[6]DealroomIncumbent ChipmakersAmazon nears a $7.5B deal for Decart, the AI startup that Anthropic walked away from and Nvidia passed on.
Read on Dealroom →
[7]explainx.aiIndustry AnalystsAmazon Nears $7.5B Decart Deal: What Is Claimed, What Is Verified
Read on explainx.ai →
More in Finance
See all →Market Outlook
Morgan Stanley Adopts Maximum Overweight on Global Equities, Raises S&P 500 Target to 8,700
7 sources
Crypto ETFs
Harvard and Dartmouth Maintain Bitcoin ETF Holdings as Fund Surges 42% in Third Quarter
4 sources
Permian Consolidation
Diamondback Energy Nears $5 Billion Acquisition of Double Eagle V in Permian Basin Expansion
6 sources
Cross-Border Investing
NSE Links GIFT City Platform to Seven European ETF Markets
5 sources
Comments
Every angle. Every day.
Get Finance stories with full source coverage and perspective breakdowns, free every day.




