Managing lots of different business units with all manner of applications is generally not an easy data management task. Making sense of all that information with data analytics can be even more challenging.
As Broadcom has grown through the acquisition of numerous software and hardware companies, the company has grappled with the challenge of integrating disparate data sources and systems.
Particularly, When the tech giant closed its $61 billion acquisition of VMware in late 2023, it faced a monumental data integration challenge. VMware operated with 1,800 disparate applications and a staggering 187,000 product SKUs. For most acquiring companies, the playbook would be familiar — implement master data management systems, build elaborate data lakes and create complex integrations to bridge the systems while gradually migrating them over several years.
Broadcom took a dramatically different approach. Instead of adding layers of complexity to manage existing systems, it wiped the slate clean. The company has also consolidated multiple data analytics tools, standardizing on just one platform from Incorta, which now works across all Broadcom’s business units, including VMware.
“VMware had a significant disparity of applications,” Alan Davidson, Broadcom CIO, told VentureBeat in an exclusive interview. “We actually deprecated every single one to consolidate to our single source management plane for the data model.”
Davidson explained that the data model is Broadcom’s source of truth for data. The goal is to consolidate and simplify the data landscape to enable more effective reporting, analytics and decision-making.
Broadcom’s consolidation approach focused on cleaning data at the source rather than transforming it later in the process. This philosophy has guided their integration strategy.
The technical process involved several critical components:
- ERP consolidation: VMware had seven enterprise resource planning (ERP) platforms. Davidson joked that they had a different one for every day of the week. Broadcom consolidated that down to one.
- SKU rationalization: VMware had 187,000 SKUs — but has now gotten that down to just 500. This radical simplification eliminated countless compatibility issues, pricing variations and support scenarios that would have otherwise required complex mapping.
- Master data restructuring: Rather than building elaborate translation layers between disparate systems, Broadcom has enforced a single master data structure. This created one authoritative source for customer records, product information, procurement data and supplier details.
- Contract and provisioning integration: A critical element was a tight integration between contracts and orders to serial numbers and provisioning. This ensures that when customers purchase software, there’s a clear line of sight from contract terms to actual usage and entitlements.
The result was a dramatically simplified architecture that eliminated the need for complex data transformation.
Standardizing on data analytics to enable enterprise insights at scale
A unified and clean source of data is the foundation of Broadcom’s data operations.
Extending that out, Davidson said his company has also chosen to unify and standardize its data analytics on a single platform. To power analytics across its massive data footprint, Broadcom chose Incorta after evaluating numerous alternatives. Currently, Broadcom has more than 17,000 internal users relying on Incorta for analytics on top of over 200 terabytes of operational data.
“Go look at any competitor in the landscape — Microsoft Power BI, Snowflake, Tableau — there’s a plethora of user interfaces (UIs), but they all come with baggage,” said Davidson. “There’s never really a complete end-to-end where I get a flexible UI, self-service capability for thousands of end users and scale in terms of usage and multiple data maps.”
Broadcom’s diverse business units, each with their own go-to-market models and data requirements, presented a significant challenge. Davidson highlighted Incorta’s ability to manage scale and link disparate data sources. Simply put, Incorta’s flexibility in segmenting data access and enabling self-service analytics has been a game-changer.
“I need to enable the segmentation, as well as link the dots together and manage the scale of the user,” Davidson explained.
Incorta’s AI and automation capabilities expand data analytics
Broadcom is now set to benefit from a series of AI and automation updates as part of Incorta’s latest product updates.
Incorta CEO and co-founder Osama Elkady explained to VentureBeat that organizations are always looking to improve access to data. To address these needs, Incorta has introduced Nexus, which provides Broadcom and other customers with generative AI capabilities.
With Nexus, users can leverage advanced natural language processing (NLP) for tasks like data cleansing, model building and connecting machine learning (ML) to gen AI. The goal is to empower Broadcom’s users with the ability to quickly and easily access the data they need, without the burden of manual data preparation or complex dashboard creation.
Applying AI with a data-first mindset
While many enterprises rush to implement gen AI without addressing underlying data quality issues, Broadcom is taking a more measured approach — focusing first on data integrity.
“The worst thing we can do is leverage AI and get inaccurate data; especially with financials and telemetry, accuracy is super important,” Davidson emphasized. “If all you’re given is garbage, all you’re going to do is get the wrong answer really quick.”
With Incorta’s Nexus, Broadcom is carefully applying AI where it adds clear value — primarily in democratizing data access.
“The whole point of the AI stuff with Incorta is transparency of data and easy access to data, without having to create your own dashboard or work hard to generate the data,” Davidson explained.
What this means for enterprise data leaders
For enterprises facing data complexity challenges — whether from acquisitions, organic growth or digital transformation initiatives — Broadcom’s approach offers several lessons:
- Question the need for complex data lakes: Instead of accepting data complexity as inevitable, consider if radical simplification at the source might be more effective.
- Standardization beats flexibility in core systems: While it requires difficult decisions, having one system of record creates fewer integration headaches than supporting multiple platforms.
- Self-service analytics requires guardrails: Broadcom’s success comes from giving users freedom within a carefully managed framework, not unlimited freedom that creates maintenance burdens.
- AI implementation requires data discipline first: Before implementing advanced AI, ensure your core data is accurate, accessible and well-structured.
As Davidson succinctly put it: “It’s hard to make things simple, and it’s easy to make things really complicated. So we try to do hard things and make them simple.”
In a technology landscape often cluttered with new tools aiming to solve integration problems, Broadcom’s approach of ruthless simplification presents a compelling alternative for enterprises undergoing significant mergers and acquisitions. By doing the difficult work of data cleaning first, they’ve created a foundation that enables, rather than inhibits, future growth.
The post Inside Broadcom’s data simplification strategy that enables 26 business units to use the same data analytics platform appeared first on Venture Beat.
