Bigabid, a Tel Aviv-based mobile demand-side platform founded in 2015, has joined AWS RTB Fabric, the Amazon Web Services infrastructure layer purpose-built for real-time bidding workloads. The move was announced on June 12, 2026, by Trevor Dyck, Head of Product for Adtech Infrastructure at AWS, via a LinkedIn post welcoming Bigabid to the growing roster of adtech companies operating on the dedicated network.

The announcement confirmed that Bigabid now runs high-frequency RTB traffic on the same private, purpose-built network environment that already hosts Amazon Ads, Affle, Azerion, GumGum, Kargo, MobileFuse, Sovrn, TripleLift, Viant, and Yieldmo. For Bigabid, whose competitive positioning depends almost entirely on the depth and speed of machine learning it applies to every bid request, the infrastructure shift addresses a constraint that had been central to its engineering roadmap.

What AWS RTB Fabric is

AWS RTB Fabric launched on October 23, 2025, offering a managed service designed specifically for real-time bidding advertising workloads. The service routes bid traffic through a dedicated, private network that consistently outperforms average public internet latency, targeting single-digit millisecond response times. Per-transaction pricing replaces the flat data transfer charges that accumulate rapidly at high queries-per-second volumes, and the service requires no upfront commitments or colocated on-premises infrastructure.

The cost reduction that Amazon claims - up to 80% compared to standard cloud networking rates - reflects the difference between paying for general-purpose internet egress and paying for a network optimised specifically for the OpenRTB request-response cycle. At the volumes that active mobile DSPs generate, that difference compounds into meaningful budget that was previously consumed by plumbing rather than by model inference or bid evaluation.

PPC Land covered the October 2025 launch in detail, noting that AWS was positioning itself not just as infrastructure for Amazon's own advertising operations but as the foundational layer for the broader programmatic ecosystem. The strategy mirrors how AWS built its cloud dominance in software-as-a-service: attract a critical mass of participants, create network effects through shared infrastructure, and generate transaction-based revenue that scales with industry growth regardless of which platform captures the most advertiser spend.

AWS RTB Fabric operates through a gateway architecture. Requester gateways function as network endpoints collocated with customer Virtual Private Clouds, routing outbound bid requests to responder gateways on the supply or demand side. The infrastructure maintains a request-response model that mirrors the OpenRTB specification without storing, processing, or examining the content of bid transactions. That separation between the networking layer and the application layer is what allows competing DSPs and SSPs to share the same infrastructure without exposing their algorithmic approaches to one another.

In May 2026, AWS added custom domain support, allowing adtech companies to migrate real-time bid traffic through Fabric using their own existing public endpoints - endpoints that in programmatic advertising often represent established, long-term traffic contracts. The CNAME-based routing mechanism means a DSP can point its existing bid endpoint to an RTB Fabric gateway without requiring any URL change from its counterparty SSPs.

Bigabid's position in mobile programmatic

Bigabid describes itself as a mobile DSP specialising in user acquisition and retargeting for mobile apps. Founded in 2015 and headquartered in Tel Aviv, Israel, the company employs between 51 and 200 people across five office locations. Its LinkedIn profile lists its specialties as technology, big data, machine learning, data analysis, real-time bidding, user acquisition, retargeting, recommendation systems, data science, DSP, and DMP capabilities.

The platform's technical differentiation rests on what it calls its Gen 2 DSP, which according to Bigabid optimises performance by analysing data in real time to target audiences at the onset of peak potential in order to ensure the highest possible lifetime value. One specific claim the company makes concerns its deep categories technology: Bigabid says its system delineates more than 1,000 app categories, versus the few dozen that most other platforms use, to eliminate low-value audiences and precisely target high-value users. That granular categorisation serves the post-IDFA environment, where the deprecation of Apple's Identifier for Advertisers removed a straightforward deterministic signal that mobile advertisers had used for years for targeting, attribution, and retargeting.

Without IDFA as a reliable common identifier, mobile DSPs have needed to substitute probabilistic and contextual signals - which puts a premium on the quality and volume of data processed per bid request. More categories, more signals, more model complexity. Each of those requirements translates directly into compute time and networking cost, which is precisely why the AWS RTB Fabric announcement matters specifically for a company like Bigabid.

The cost and compute argument

Amit Attias, CTO of Bigabid, was quoted directly in both the LinkedIn announcement post and the AWS RTB Fabric customer page. The core framing of his statement identifies the networking cost as a ceiling on ML performance:

"Bigabid's edge is the depth of ML we run on every bid request - but that only works if we can evaluate enough opportunities, fast enough, at a cost that pencils out. AWS RTB Fabric removed that ceiling: we saw over 80% savings on networking costs versus the public internet, and reinvested that directly into model compute and additional QPS. It's rare to find an infrastructure decision that's both an immediate cost win and a long-term performance unlock for our advertisers."

The key phrase is "additional QPS" - queries per second. In real-time bidding, a DSP receives bid requests from exchanges and SSPs and must evaluate each one, decide whether to bid, set a bid price, and return a response, typically within 100-150 milliseconds. The number of bid requests a DSP can evaluate per second determines the size of the auction universe it can participate in. A DSP that can double its QPS - either by reducing per-request compute cost or by increasing infrastructure budget - effectively doubles the number of auction opportunities it evaluates. That scale has direct implications for campaign performance: more auctions evaluated means more opportunities to find the precise user at the precise moment a campaign targets.

The claim of over 80% networking cost savings aligns with what multiple other AWS RTB Fabric participants have reported. TripleLift, the creative SSP, described a 15-20% reduction in latency, a 30% increase in ad spend, and a 30% increase in win volume over the public web, alongside an over 80% reduction in networking costs compared to public pricing. Yieldmo, the advertising platform, reported more than an 80% reduction in timeouts and higher bid rates. Azerion reported a reduction in timeout and connection errors of over 50%. Viant Technology, which trades on the Nasdaq under the ticker DSP, reported over 80% savings in networking costs on average with early access to AWS RTB Fabric.

The pattern across those reports is consistent. Networking overhead on the public internet - packet loss, variable latency, congestion at peak load - creates a baseline inefficiency that compounds at RTB scale. A dedicated private network eliminates most of that baseline, which simultaneously reduces costs and improves bid response rates, since fewer responses are dropped or returned after the timeout window.

The broader context of RTB Fabric's partner roster

Bigabid joins a network that spans both the buy side and the sell side of programmatic advertising. Affle is a global technology company enabling AI-led solutions in mobile advertising. According to Affle's chief architect and technology officer Charles Yong, the company has achieved up to 80% higher network efficiency by leveraging AWS RTB Fabric, resulting in lower latency, improved data enrichment, and reduced cloud costs.

Amazon DSP, which is also on the platform, is itself a buyer on AWS-hosted infrastructure - a configuration that underlines the dual role AWS plays as both infrastructure provider and ad platform operator. Dr. Neal Richter, Director of Amazon DSP and Chair of IAB Tech Lab, described the service as addressing a foundational need in the programmatic ecosystem: "The programmatic ecosystem has long needed a neutral, scalable network for buyers and sellers to interoperate without costly technical overhead."

Amazon Publisher Services, which operates at multimillion queries-per-second scale across global RTB workloads, is also on the platform. Scott Siegler, Director of Amazon Publisher Services, described the prior gap in industry infrastructure: "The industry has been waiting for a dedicated cloud service that can handle this level of performance, reliability, and partner complexity - at a price point that fits the reality of adtech margins."

GumGum, the contextual SSP, noted that the cost efficiencies allow it to send more bid requests to DSP partners, increasing revenue for both GumGum and its partners. Sovrn, which serves thousands of customers across 80,000-plus websites, mobile apps, and CTV channels and reaches over 500 million active consumers across more than 40 billion page views per day, described the service as making the "promise of lowering infrastructure costs and improving efficiency for RTB workloads in the cloud" a reality. Kargo, the brand advertising company, described the service as representing "a much-needed shift toward a more open, interoperable, and efficient adtech infrastructure."

The fact that competing platforms - DSPs that are bidding against one another for the same impressions, SSPs that are competing for the same demand - operate on shared infrastructure is not unusual in cloud computing terms, but it is notable in the context of an industry that has historically treated infrastructure as proprietary. The RTB Fabric model separates the networking layer from the algorithmic and data layers that define competitive advantage, making the former a commodity while leaving the latter proprietary.

Why this matters for performance advertisers

For performance advertisers running mobile user acquisition or retargeting campaigns, the specific infrastructure decisions made by their DSP partner are rarely visible. Campaign managers see bid win rates, cost-per-install figures, lifetime value metrics, and return on ad spend. They do not typically see how many bid opportunities were evaluated, how many were dropped due to timeout, or what fraction of the infrastructure budget went to networking rather than model inference.

The significance of the Bigabid-AWS RTB Fabric integration is that it changes the internal allocation of fixed infrastructure budget - and that change, if Attias's framing is accurate, flows directly into advertiser performance rather than into margin improvement. Reinvesting networking savings into model compute and additional QPS is a specific claim about where the efficiency gain goes. Whether it holds in practice depends on execution, but the mechanism is logical: lower networking cost per bid request means that the same budget processes more bid requests, which means more auction participation, which means more opportunities to find high-value users.

Programmatic advertising reached $162.4 billion in the United States in 2025, up 20.5% year-over-year according to the IAB, adding $27.6 billion in new spend over the course of the year. Mobile remains a significant component of that total. The infrastructure layer underneath that spend is, increasingly, AWS.

Amazon's advertising business crossed $70 billion on a trailing twelve-month basis in early 2026, and AWS RTB Fabric is now running in production across six regions: US East (N. Virginia), US West (Oregon), Asia Pacific (Singapore), Asia Pacific (Tokyo), Europe (Frankfurt), and Europe (Ireland). Those six regions cover the primary programmatic advertising markets in North America, Asia, and Europe. Transaction-based pricing means AWS benefits from the growth of programmatic advertising volumes regardless of which platforms or advertisers capture individual impressions.

What Bigabid's integration signals

Bigabid is the only mobile-focused pure-play performance DSP currently named on the AWS RTB Fabric customer page. The others on that page represent a mix of mobile and desktop inventory on the sell side (Azerion, GumGum, Kargo, MobileFuse, Sovrn, TripleLift, Yieldmo), a CTV-focused DSP (Viant), a mobile-focused demand platform (Affle), and Amazon's own buy-side and publisher-services operations.

That positioning places Bigabid in a distinct category within the RTB Fabric ecosystem: a mobile user acquisition and retargeting platform whose competitive logic is ML depth rather than inventory breadth or brand advertising formats. Trevor Dyck's LinkedIn announcement described the company as one that "helps performance advertisers build smarter, more accountable campaigns for user acquisition and retargeting" and noted that "by moving high-frequency RTB traffic onto the purpose-built network environment of RTB Fabric, the Bigabid team can run deeper ML on more bid requests while spending less on the infrastructure underneath, so their engineering focus stays on models and advertiser performance rather than systems overhead."

The distinction between engineering focus on models versus engineering focus on systems overhead is not incidental. At a company of 51-200 employees operating across five locations, engineering capacity is a constrained resource. Infrastructure that absorbs significant engineering time - maintaining networking configurations, monitoring public internet performance, managing egress costs - competes with time spent improving model accuracy, expanding training data, or developing new optimisation features. The RTB Fabric integration is as much an engineering allocation decision as it is a networking cost decision.

Timeline

Summary

Who: Bigabid, a mobile demand-side platform founded in 2015 and headquartered in Tel Aviv, Israel, employing between 51 and 200 people, joined AWS RTB Fabric. The announcement was made by Trevor Dyck, Head of Product for Adtech Infrastructure at Amazon Web Services, on June 12, 2026. Amit Attias, CTO of Bigabid, provided the quantified performance statement.

What: Bigabid migrated its high-frequency real-time bidding traffic to AWS RTB Fabric, a purpose-built private network for RTB workloads that Amazon launched in October 2025. The integration produced over 80% savings on networking costs compared to the public internet, with those savings reinvested into machine learning model compute and additional queries-per-second capacity. Bigabid joins Affle, Amazon Ads, Azerion, GumGum, Kargo, MobileFuse, Sovrn, TripleLift, Viant, and Yieldmo on the platform.

When: The announcement was published on June 12, 2026, via a LinkedIn post from AWS. AWS RTB Fabric itself launched on October 23, 2025, and has been running in production since that date across six AWS regions.

Where: Bigabid operates from Tel Aviv, Israel, across five office locations globally. AWS RTB Fabric operates across six AWS regions: US East (N. Virginia), US West (Oregon), Asia Pacific (Singapore), Asia Pacific (Tokyo), Europe (Frankfurt), and Europe (Ireland).

Why: Bigabid's competitive approach depends on running deep machine learning on every bid request. High-frequency RTB traffic over the public internet carries networking costs that constrain how many bid opportunities a DSP can evaluate and how much compute it can apply per evaluation. By moving to RTB Fabric's private network - which charges per transaction rather than per gigabyte of egress - Bigabid reduced its networking cost floor and freed budget to expand model compute and bid throughput. For performance advertisers running mobile user acquisition and retargeting campaigns, the implication is greater auction participation and deeper per-impression evaluation, which is the basis on which mobile DSPs compete for high-value users.