Oracle building
Photo: Oracle

The man behind Larry Ellison’s health care gamble

Protocol Enterprise

Hello, and welcome to Protocol Enterprise! Today: why Cerner CEO David Feinberg could ultimately decide whether Oracle’s big acquisition pays off, Databricks starts a land war in open source and Intel continues to emphasize its software investments.

David Feinberg’s magic touch

With Oracle’s $28.3 billion acquisition of Cerner officially complete as of today, Larry Ellison can now count one of the most-well known health care tech brands as a member of his fiefdom.

And Ellison has a powerful tool at his disposal to help Oracle in the battle with Microsoft and other key rivals in the quest to bring the cloud revolution to the health sector: Cerner CEO David Feinberg.

  • The former CEO of UCLA Health, Feinberg has deep ties in health care and is viewed as a trusted partner, a key quality for a sector that is notoriously wary of making big tech gambles.
  • In his nearly three years as vice president of Google Health, Feinberg struck a slew of lucrative partnerships with companies including Epic and HCA Healthcare. Some were more successful than others.
  • “The magic of David is he helps open doors,” said SVB Securities senior research analyst Stephanie Davis. The health care industry is “very relationship-driven. Folks are betting their career on something a bit more innovative. Because of that, it becomes more important to have someone” like Feinberg, she added.

Oracle has a reputation of quickly infusing acquired products into its mammoth sales operation. However, in a sign Oracle might be learning from past mergers, it’s looking likely that Oracle will let Cerner operate separately from the mothership — at least initially.

  • The company has a spotty track record in terms of being able to keep founders or CEOs of acquired companies on its payroll.
  • In 25 deals that Oracle struck over the past decade, just nine of the founders or CEOs at the time of the merger still remain at the company, according to a Protocol analysis. Oracle declined to comment.
  • The company, however, isn’t the only one that has struggled to keep influential executives post-merger. Former Segment CEO Peter Reinhardt, for example, left Twilio shortly after the one-year anniversary of the $3.2 billion deal.
  • Conversely, Bret Taylor, who joined Salesforce in 2016 after the Quip merger, is now the company’s co-CEO.

Feinberg is particularly important as Oracle looks to overcome a key hurdle to penetrating the health care industry: its reputation.

  • Historically, Oracle has operated one of the strongest sales teams in enterprise tech. But after years of questionable behavior and aggressive sales and pricing tactics, buyers are wary of aligning too closely with Big Red.
  • Feinberg can likely help mitigate some of the concerns that exist around jumping on the Oracle bandwagon, especially for health care companies that are just now beginning to embrace the cloud and might be nervous about migrating too quickly.

Whether that is enough to succeed in a sector where so many others have failed remains to be seen, especially after Microsoft’s $16 billion acquisition of Nuance. But it’s hard to bet against Oracle with someone like Feinberg in the company’s corner.

— Joe Williams (email | twitter)


Fewer than half of executives (44%) see better communication with customers as a benefit of digitizing AR. Meanwhile, 72% state that their AR department isn't customer-oriented enough, implying that executives understand the need for customer-oriented AR departments, but aren't aware that they can close that gap as part of their AR digitization project.

Learn more

Delta blues

Database wranglers were already questioning Delta Lake’s open-source cred. So it was only a matter of time before a wonky “nerd war” over the Databricks-led open-source project — and a debate common in other tech circles — came to database talks on LinkedIn.

There’s confusion around which parts of Delta Lake are open source and which aren’t. That can have important ramifications for users down the line.

  • Data engineers say Databricks puts up roadblocks to Delta’s full capabilities, forcing users to pay for access to its full speed and breadth of features.
  • Choosing between the two can have a real business impact, said Billy Bosworth, CEO of Dremio, whose company has highlighted its use of Iceberg in its own products. “It's an architectural decision of the type where you live with it for about a decade or more when you make it. So it's a very critical point in the architecture,” he said.
  • Ali Ghodsi, co-founder and CEO of Databricks, pushed back: “Our platform documentation explains which performance features are only available on Databricks,” he told Protocol.

Amid squabbles over Delta Lake, momentum behind rival project Iceberg grows.

  • Along with adoption by Dremio and Snowflake, AWS used Iceberg to build its Athena query service, which was made broadly available in April.
  • Google is backing Iceberg, too: “We are supporting Iceberg first with BigLake because that’s the demand that we see on GCP,” said Gerrit Kazmaier, vice president for Database, Data Analytics and Looker at Google.
  • But Iceberg hasn’t impressed Microsoft customers, said James Serra, a data and AI solution architect at Microsoft: “When it comes to Iceberg, I honestly haven’t seen any customers at all using it.”

Purists say Delta Lake is not open source in spirit.

  • Unlike Iceberg, Delta critics say the project has not fostered a robust, diverse and collaborative community typical of true open-source projects.
  • Databricks employees wield undue control over decisions to make adjustments to its code without public review, Iceberg supporters say.
  • “Thousands of our customers — non-Databricks employees — are active in the community,” said Denny Lee, head of Developer Relations at Databricks, of the Delta community on GitHub.

With Spark, the co-founders of Databricks have a history of monetizing the open-source community.

  • To launch Databricks, co-founders packaged improved features for Spark — the popular open-source project they started — into a better-performing, paid product.
  • “You can say they’re trying to do the same thing with Delta Lake,” Microsoft’s Serra said.
  • “This seems to me like slightly disingenuous behavior,” said Armon Petrossian, CEO of data transformation and analytics company Coalesce.

Check out the full story of this heated Delta Lake open-source discussion and why it matters to businesses.

— Kate Kaye (email| twitter)

In AI, it’s the software — not the chips

Twenty-year Intel veteran Sandra Rivera presides over Intel's ever-important data center and AI group at the heart of its turnaround efforts. In a recent interview at Intel Vision 2022, Rivera discussed the company’s aim to execute the plan outlined by CEO Pat Gelsinger last year, and how it is thinking about its data center AI efforts.

For years, Intel has insisted that the CPU was sufficient to handle all the tasks in the data center. That’s changed with the company’s launch of infrastructure and graphics chips, among other things. Why?

Pat has accelerated everything, but the idea of heterogeneous compute, we've had that idea for quite some time. I think everyone recognizes that particularly with the growth of data being both structured and unstructured, dense data, sparse data — it is not a one-size-fits-all.

If all you’re doing is deep-learning training, then actually even a GPU is not your best choice. You want an AI accelerator, but you will only deploy that at scale if the software environment is one that is familiar, if you’re able to access it through PyTorch and TensorFlow.

With the computing horsepower required for complex AI rising considerably faster than the current rate of silicon advancements, how important is software to running workloads related to AI?

Hardware gives a certain process [advancement] with each generation of technology, architectural and design engineering improvements from a hardware perspective. But the real unlock of that hardware is in the software — it will get you many more multiples than you’re going to get in hardware or process technology.

When we look at the Ice Lake [Intel server chips] generation with Deep Learning Boost, and then what we're bringing forward with Sapphire Rapids and Advanced Matrix Extensions, we see a 30X improvement in running something like human-genome sequencing data set algorithm.

— Max A. Cherney (email | twitter)

Around the enterprise

MongoDB introduced several new features for its databaseat an event Tuesday, including support for encrypted data queries.

GitLab stock spiked after it reported earnings Monday well above Wall Street expectations, including a 75% jump in revenue.

Apple’s Xcode CI/CD service, one of the first enterprise cloud developer services it has ever released, is now generally available.

Figma suffered a widespread, multihour outagethat as of publish time was still ongoing.


A resounding 96% of respondents claimed that there is work to do in digitizing their AR departments, yet 60% agreed that their AR departments haven’t been prioritized as much as other departments for digitization. At a time when the importance of securing cash flow is higher than ever, many businesses are not putting enough focus on it.

Learn more

Thanks for reading — see you tomorrow!

Recent Issues