Rakuten's Technology Stack in 2026: From E-Commerce to OpenRAN to Symphony
Rakuten is one of the most interesting and underestimated technology stories of the decade. A deep look at the stack and the strategic bet, in 2026.
Rakuten started as an online marketplace in 1997 — a Japanese answer to early Amazon. Through the 2000s and 2010s it diversified aggressively into banking, brokerage, credit cards, travel, content, sports sponsorship, and a long tail of digital services. The strategic logic — a “Rakuten Ecosystem” where points and payments flowed between businesses — was novel and substantially successful. Then in 2019, Mickey Mikitani announced something that read at the time as wildly out-of-character: Rakuten would build its own mobile network from scratch, as a fourth carrier in Japan, using a fully software-defined Open RAN architecture.
Seven years later, in 2026, the mobile bet is the most interesting thing Rakuten is doing — and the technology platform behind it (Symphony) is a real product being sold to telcos globally. The financial picture is still mixed (the mobile build has been more expensive than projected, the carrier business is losing money, the broader group is in transition), but the engineering work is genuinely consequential.
I want to walk through the actual technology stack, because it is more interesting than the headlines.

The e-commerce and fintech base#
The legacy Rakuten businesses — Rakuten Ichiba (marketplace), Rakuten Card (credit), Rakuten Bank (banking), Rakuten Securities (brokerage), Rakuten Insurance, Rakuten Travel — collectively serve 100M+ Japanese members. The technology base is heterogeneous (the businesses were partly acquired, partly built), running on a mix of on-prem data centers, AWS, and increasingly Rakuten’s own Symphony cloud platform.
The user-side glue is the Rakuten Super Point loyalty system — a unified points balance earned and redeemable across the businesses. This is the moat. A user who earns points buying groceries on Ichiba can redeem them paying for a Rakuten Mobile bill or buying mutual funds through Securities. The points system is conceptually simple, technically non-trivial across heterogeneous business systems, and operationally critical.
Rakuten Pay — the QR payment offering — competes with PayPay and LINE Pay in the Japanese payments landscape. It is the third-largest in the segment, anchored by the broader Rakuten Ecosystem rather than by independent merchant acquisition.
Rakuten Mobile and the OpenRAN bet#
The mobile carrier — Rakuten Mobile — entered the Japanese market as a fourth carrier in April 2020 (alongside NTT Docomo, KDDI au, and SoftBank). The strategic premise: build the network from scratch with a cloud-native, fully software-defined Open RAN architecture, rather than buying integrated vendor equipment from Ericsson, Nokia, Samsung, or Huawei. The cost claim was that fully-virtualized infrastructure would be 30-40% cheaper to build and 40-50% cheaper to operate than traditional carrier infrastructure.
The reality has been more mixed.
The technology delivered on much of what was promised. Rakuten built a working 4G network in three years and a 5G overlay in four. The fully-virtualized architecture is operational. The cost-per-bit-of-capacity claims hold up directionally even if the absolute numbers have been higher than initial projections.
The carrier business has been financially difficult. The capex was higher than projected — building national coverage in Japan, with its dense urban and mountainous rural mix, is hard. The price war with the incumbents (Rakuten launched with an aggressively low price point) compressed margins. The subscriber acquisition has been substantial (8M+ subscribers as of 2026) but below the original ambition.
The platform — Symphony — has emerged as a serious product. This is the part that matters strategically.
Rakuten Symphony#
Symphony is the productized version of the technology platform Rakuten built for its own mobile network, now sold to other carriers. The components:
Symphony Cloud — a cloud-native infrastructure platform optimized for telco workloads. Kubernetes-based, with extensive customization for the deterministic latency and high-availability requirements of carrier-grade applications.
Symphony RAN — the Open RAN distributed unit, central unit, and management software. Replaces the integrated vendor RAN equipment in traditional carrier architectures.
Symphony Mobile Core — a 5G core network implementation. Compatible with the 3GPP specs but with cloud-native architecture choices that differ from traditional vendor cores.
Symphony OSS/BSS — the operations support and business support systems. The plumbing that makes a carrier work — customer management, billing, network management, service activation.
Symphony AI — increasingly, AI products layered on the platform for network optimization, customer churn prediction, and operational automation.
The customers in 2026 include 1&1 in Germany (a major commercial customer, building a new mobile network on Symphony), Dish Network in the US (also building OpenRAN from scratch), and a growing list of smaller carriers in Europe, the Americas, and Asia. The pitch — cloud-native, vendor-neutral, lower cost of operation — is increasingly credible as the deployed customer base grows.
The strategic implication is that Rakuten has positioned itself as the platform play in OpenRAN, similar to how Microsoft positioned itself in cloud or Stripe in payments. If OpenRAN succeeds as a category — and the trajectory in 2024-2026 has been increasingly positive — Symphony is the largest pure-play OpenRAN platform.
The AI and data infrastructure layer#
Across the Rakuten Group, the AI and data infrastructure is being increasingly unified. The acquisition of Cuetzpalin (a small AI lab) and the establishment of Rakuten Institute of Technology have produced an internal foundation-model effort plus enterprise AI products.
Rakuten AI — the brand for the group’s AI offerings, including conversational interfaces in the apps, AI-augmented customer service, and developer tooling for product teams.
The unified customer data platform spans the various Rakuten businesses with strict consent controls (Rakuten’s privacy posture is influenced by both Japanese APPI and the EU GDPR via the European subsidiaries).
Internal LLM deployment — Rakuten has deployed Claude, GPT, and Sarvam models internally for various engineering and business workflows, with the AI gateway pattern that lets workload-specific routing.
The strategic question for 2026#
The strategic question facing Rakuten in 2026 is whether the diversified portfolio — e-commerce, fintech, mobile, content — coheres as a single asset or whether the parts are worth more separated. The market view is mixed. The mobile carrier business has been a financial drag; the Symphony platform is a credible growth bet; the e-commerce and fintech businesses are mature but cash-generative.
The recent corporate actions — Symphony IPO discussion, the consolidation of the various financial services entities, the careful management of the carrier business toward operational sustainability — suggest a path forward that involves continued portfolio management rather than wholesale separation.
For technology observers, the most interesting question is whether Symphony can become the AWS-of-telco — a platform that other carriers run on, with Rakuten’s own mobile business as the largest first customer. The pattern has worked in cloud (AWS started as Amazon’s internal infrastructure), in payments (Stripe was famously Patrick and John’s project before it was a platform), in observability (Splunk, Datadog all started internally). Whether it can work in telco — an industry with very different procurement, regulatory, and operational dynamics — is the open question.
What outsiders can learn#
Three lessons from Rakuten that builders outside Japan should internalize.
1. The ecosystem effect compounds. Rakuten’s points system — earning and redeeming across businesses — is the kind of network effect that takes a decade to build but produces durable customer lock-in. Companies in other markets attempting similar plays (Tata’s Neu in India, Naver/Kakao in Korea) are working from this playbook.
2. Vertical integration is unfashionable but powerful. Building your own mobile network sounds insane in 2019 and looks shrewd in 2026 — if the technology bet pays off. Rakuten took a real bet and is now positioned to monetize the platform regardless of whether the carrier business is financially optimal.
3. The platform business inside the operating business. AWS was Amazon’s internal infrastructure first. Stripe was the founders’ internal payment infrastructure first. Symphony was Rakuten’s internal mobile infrastructure first. The pattern — productize what you built for yourself — is increasingly the dominant pattern for new platform businesses.
Where pdpspectra fits#
We do not work directly with Rakuten, but we work with companies in adjacent positions — large groups with diversified businesses considering platform plays, telco operators evaluating Open RAN vendors, fintechs building loyalty and points infrastructure. Our architecture practice does this kind of strategic technology work.
Related reading: the Japan payments landscape post, the Japan cloud adoption post, and the AI gateway pattern post.
Rakuten is the most under-studied Japanese tech story. Talk to our team about your platform strategy.