Welcome to Runtime! Today: A look back at some of the most notable developments from the past year and the last Runtime Roundup of 2024.

(Was this email forwarded to you? Sign up here to get Runtime each week.)

The way we were

Given its relentless focus on creating the future, the tech industry tends to be really bad at reflecting on past events — both good and bad — that frame those future outcomes. That's why you often get a flurry of predictions around the end of the calendar year from publications such as this one that, like most of us at the end of a long year, are simply looking to run out the clock.

But his year marked a turning point for enterprise tech as spending recovered thanks to hopeful investments in generative AI technology and the economy stabilized following years of rising interest rates and supply-chain disruption. While no one knows what lies ahead, here are five things we thought summed up a pivotal year.

The rise of agentic AI: The lofty rhetoric surrounding generative AI in 2023 came back down to earth this year — somewhat — thanks to scaling issues and deployment problems. But the release of new AI models allowed vendors to shift the goalposts and proclaim agentic AI as the revolution we were all promised after the release of ChatGPT in late 2022.

AI agents could allow companies to build much more sophisticated customer-service applications and automate business processes with greater accuracy and efficiency than ever before. With nearly every enterprise software vendor announcing an agentic AI strategy in the second half of 2024 — led by Salesforce CEO Marc Benioff, who likes a bandwagon — we'll see if tech buyers bought the pitch by the middle of next year.

The collapse of Intel: One of the most important companies in the history of information technology was already in trouble when prodigal son Pat Gelsinger returned in 2021 to try and clean up the mess, but it all came crashing down in 2024. The last straw appeared to be its second-quarter earnings report, in which it missed expectations, lowered guidance, and announced plans to lay off 15% of its workforce.

Gelsinger was forced out in early December, and the company still does not seem to have a plan for regaining momentum in both chip manufacturing and design. Intel was a crucial piece of the Biden administration's hopes to restore American chip manufacturing, and despite receiving billions of taxpayer dollars this year the company remains on very shaky ground.

Snowflake's open shift: Snowflake helped break ground for the generative AI boom by giving businesses around the world a powerful modern data warehouse born around cloud computing principles, and customers poured reams of data (and money) into its products over the last decade. However, Snowflake charged customers for storing that data in its proprietary formats, which wasn't a huge problem when money was flying around yet quickly became a problem as budgets tightened up.

Databricks, on the other hand, allowed customers to store data wherever they liked in open formats, which was cheaper and more flexible as AI demand exploded. Snowflake's decision to embrace open formats this year may have been forced by its rival, but sets up a future in which data analysis customers can store their data in cheap buckets and mix and match processing engines as they see fit.

Microsoft's security reset: Like its longtime partner Intel, Microsoft's cloud security issues were a cause for concern among enterprise buyers for several years, but this year its most important customer decided it had seen enough. In a scathing report released this past April, the U.S. government declared that several incidents in recent years "point to a failure of Microsoft’s organizational controls and governance, and of its corporate culture around security."

To its credit, Microsoft owned up to the problem and vowed to elevate security as a corporate value through a series of changes outlined in May. Executives are now compensated based on their commitment to security, and the company doesn't appear to have suffered a major breach or incident since then.

Matt Garman's time: After Adam Selipsky trotted his way through a race to catch up in the generative AI boom, AWS passed the baton to longtime sales and operational executive Matt Garman, who became just the third CEO in AWS history in May. A company as large as AWS doesn't turn on a dime, but Garman's impact began to show up throughout the summer as AWS shut down several older and languishing services that in past years, it would have kept supporting no matter the cost or effort.

And at re:Invent, Garman struck a balance between AWS's core infrastructure services and the generative AI opportunity that seemed far more in harmony after two years of relentless focus on the latter. However, Amazon CEO Andy Jassy hovered over the event, snatching several precious minutes of the Tuesday keynote to personally unveil Amazon's new AI models and recording nightly video recaps of each day's announcements, which intentionally or not suggested that Garman might be on a short leash after the announcement of his hiring implied he was passed over three years ago in favor of a "more seasoned CEO."

The Runtime roundup

A limited version of GitHub Copilot is now available for free directly within Microsoft's VS Code IDE, with the option of using OpenAI's GPT-4o or Anthropic's Claude 3.5 Sonnet model.

Cisco announced plans to acquire SnapAttack, a threat detection company it plans to integrate within Splunk.

ScllyaDB is dropping the open-core business model and will switch to producing a "single release stream" called ScllyaDB Enterprise under a source-available license.

Thanks for reading — Runtime is off until January 6th, 2025 — see you in the new year!