AWS Re:Invent - Quick thoughts

Attending AWS re:Invent this year was a deeply emotional experience for me. As an early leader at AWS, I hadn't attended since the event launched the year I left for roles at Microsoft and DellEMC. Witnessing the scale AWS has achieved—and reconnecting with old colleagues—was both inspiring and humbling.

The keynotes stood out as highlights, especially those by Peter Desantis and Matt Garman. Peter, my first hiring manager at Amazon and now an SVP, spoke powerfully about Amazon's culture of innovation and showcased incredible advancements, like their Ultra server delivering 83.2 PFLOPS of compute power. On the day Peter welcomed me to Amazon, I doubt either of us imagined Amazon building its own cutting-edge chips and servers, let alone challenging Nvidia on AI training or delivering low-latency inference engines. It's a testament to the core principles that drive Amazon's rapid innovation.

Matt Garman's keynote was equally engaging, with a strong focus on inference as a priority. If AWS combines hardware advancements with real software improvements for diverse inference workloads, they'll have a winning formula. Innovations like the Firefly connector solving network speed and reliability challenges for 10p10u racks show the kind of bottom-up ingenuity AWS excels at.

However, the pace of top-down innovation—like in storage and data—has been less impressive. S3, the backbone of so many data infrastructures (Databricks, Netflix, and others), has seen little evolution in features over the years. The recent launch of S3-Tables is promising but feels overdue and limited compared to other table abstractions built atop S3. AWS has missed opportunities to lead here, despite owning the "storage for the internet." (refer to the rather biased but true-to-a-large-extent note from Vinoth Chandar)

Finally, Werner Vogels' talk was insightful, though perhaps overly technical for a general keynote audience. Through my experiences at various organizations post-AWS, I have come to appreciate that while the core principles articulated by Peter DeSantis empower Amazon to move swiftly, true differentiation is largely a reflection of strong leadership. Under the stewardship of Peter DeSantis and Matt Garman, EC2 and AWS infrastructure have experienced remarkable growth. What began as EC2 on commodity hardware, manufactured by third-party server makers and supported by open-source virtualization software, has transformed into a sophisticated and innovative offering. Enhancing leadership across other domains could allow AWS to achieve similar levels of innovation and unlock billions of dollars in market value.

I have had the privilege of working closely with all three leaders. Peter Desantis played a pivotal role in helping me launch Amazon’s first messaging product/service. Wener Vogels provided valuable insights and guidance during the initial design phases. Meanwhile, Matt Garman supported my transition into a General Manager role, assisting on the business front. I remain highly optimistic that AWS will continue to achieve unprecedented success under their visionary leadership.