In one of the most detailed blog posts yet [3,200 words], Netflix describes the custom Origin Server it built for live. The sentence I like most states that Netflix took a platform originally built for VOD and simply “extended nginx’s proxy-caching functionality to address live-specific needs.” This goes against those who claim that Netflix can’t deliver live streaming at scale because it “wasn’t built for live,” and of course, contradicts all the public viewership data, with Netflix having successfully streamed live events with an AMA number in the tens of millions. There is so much great info in the post, so here’s just a few things I wanted to highlight:

➡️ Leveraging Netflix’s microservice platform priority rate limiting feature, the origin prioritizes live edge traffic over DVR traffic during periods of high load on the storage platform. To mitigate traffic surges, TTL cache control is used alongside priority rate limiting. When low-priority traffic is impacted, the origin instructs Open Connect to slow down and cache identical requests for 5 seconds by setting max-age = 5s, and returns an HTTP 503 error code. This strategy effectively dampens traffic surges by preventing repeated requests to the origin within that 5-second window.

➡️ Netflix’s combination of cache and highly available storage has met the demanding needs of its Live Origin for over a year, and the solution they built was significantly more expensive, but minimizing cost was not a key objective, and low latency with high availability was.

➡️ Netflix uses failover orchestrated at the server-side to reduce client complexity, with resilience achieved through redundant regional live streaming pipelines and implements epoch locking at the cloud encoder, which enables the origin to select a segment from either encoding pipeline.

➡️ Netflix’s redundant cloud streaming pipelines operate independently, encompassing distinct cloud regions, contribution feeds, encoder, and packager deployments to substantially mitigate the probability of simultaneous defective segments across the dual pipelines.

➡️ Millisecond grain caching was added to nginx to enhance the standard HTTP Cache Control, which only works at second granularity, a long time when segments are generated every 2 seconds.

➡️ Netflix’s enhanced cache invalidation and origin masking enable live streaming operations to hide known problematic segments from streaming clients once the bad segments are detected, protecting millions of streaming clients during the DVR playback window.

You can read the entire Netflix blog post here.

Stats Perform acquired the assets of the ultra-low-latency tech company Phenix Real Time Solutions for approximately $7 million, including 21 patents. Stats Perform also rehired most of the former employees, with approximately nine coming over. In June of 2025, VC firm KB Partners put the assets of Phenix Real Time Solutions up for sale after the company continued to lose money, to the tune of almost $10 million last year, on sales of about $4 million.

No vendor can survive by selling a low/ultra-low latency solution as a stand-alone offering, as it must be sold as part of a larger video platform. This is exactly why the asset acquisiton makes sense for Stats Perform, which offers data collection and predictive analysis services tied to fantasy sports and sports betting. Ultra-low-latency tech can also help prevent piracy, a topic Stats Perform is discussing with leagues, OTT platforms and rights holders.

In the betting world, ultra-low-latency tech is needed to improve fan engagement, and Stats Perform has customers including the Premier League, WTA, FIBA, La Liga, and Serie A. Stats Perform is also FIFA’s first-ever official global distributor for betting data and live streaming rights, so it will be interesting to watch if FIFA wants to leverage any of the tech within their services. I can also think of how the ULL tech might be used to keep video in sync within a multiview feed.

Stats Perform tells me that while Phenix’s WebRTC-based tech works really well, Phenix never built the platform to scale or be cost-effective. Phenix never shared many details about its costs, but it once admitted that its platform was not designed to support more than 500,000 concurrent streams and that the overhead was expensive, even when it wasn’t being used.

Stats Perform has relaunched the Phenix tech within its platform, having already cut operating costs in half and continues to scale the tech, further reducing costs. This is exactly how ULL technology should be deployed, as a feature of a larger platform, not as a stand-alone service. Stats Perform will be well-positioned to monetize the tech across its entire platform for teams, leagues, broadcasters, betting companies, and social media. Stats Perform dominates its industry with an annual run rate of more than $550 million, nearly 2,800 employees and owns more than 560 patents.

I say this at the end of every year, but the last twelve months have seen an incredible amount of news and deals in the streaming industry, making it another busy year, with plenty to discuss. I want to thank all those who read my content across LinkedIn and my blog, contributed to the discussions and listened to my podcast. For me, 2025 marks 30 years since I worked on my first webcast, with the Macintosh New York Music Festival in NYC, and my role as an Apple Systems Engineer, which led to where I am today.

I will end 2025 with just over 10 million impressions on LinkedIn, becoming a Top Voice, reaching more than 1.7 million members, along with 2.8 million impressions on my blog, with the podcast nearing 100,000 downloads since launch – all while keeping my content free. While many have tried to convince me to monetize my content via paywalls, charging for newsletters or selling limited access, that’s not why I do this. Not everyone is driven by the need to monetize content, and the currency I focus on in the industry isn’t money, but something more valuable – trust. With readers, vendors, broadcasters, OTT platforms, sports leagues and many others, trust and relationships are my focus.

The industry has taken great care of me over the past three decades, for which I am grateful. I feel a sense of responsibility to help grow the industry by doing my best to inform, educate, and empower others, connecting them with those who are more intelligent than I am and separating facts from opinions. Good leaders invest in people, not ideas, and I am always trying to give back and invest in those in our industry.

Some think I am crazy to have published my cell phone number publicly for 25 years and, on average, reply to all calls and emails within 24 hours. But as a result, that’s where some of my best information comes from, the many people in the industry who are far smarter than I am, giving me the inside details on what’s really taking place. While some don’t like how direct I am in my speech and writing, that’s the number one request I hear from readers: to tell it like it is and provide the facts, without sugar-coating anything. You can be direct and also be professional at the same time.

For all those who have contributed to helping me tell stories, pointed out my spelling mistakes, sent me messages thanking me for my posts, and educated me on topics they know better than I do – I thank you. Next year is going to be a busy one. Get some rest, spend time with your family, catch up on some good video content and get ready for what looks to be a ridiculous amount of streaming news and deals in 2026. Happy Holidays! 🎄 📺

Since I write about vendors, I want to disclose that, effective December 2nd, I have become a minority shareholder in Hydrolix through a personal investment in the company. This is only the second time I have invested in a company in the streaming industry, the other being my seed investment in Datazoom, which I disclosed in a blog post in 2018.

Hydrolix really came onto my radar about 18 months ago when they started hiring a lot of smart people across the industry I respect, and I began to hear their name mentioned by many large OTT platforms and broadcasters, impressed with their platform. I’ve been saying for years that the ingestion, processing, and storage of user data tied to the streaming media workflow are the industry’s most significant problems. It’s not video compression, delivery or playback. Today, many large companies process over 1 billion transactions per day and ingesting and analyzing data at that scale has historically been impossible.

If you can’t manage it, you can’t measure it. And that’s the point of Hydrolix’s real-time data platform, which helps businesses manage and analyze massive volumes of data, enabling them to make informed decisions about their business. The challenges media and entertainment customers face with the volume of data they collect across siloed systems and multiple dashboards are real problems. Many store only 1% of the data they collect and discard the rest due to storage costs, or store it in cold storage, making it inaccessible in real time. These are all problems Hydrolix is working to solve for companies such as FOX, DAZN, and TF1. Further, through their OEM with Akamai, they’re delivering ingest and insights for over 600+ other major media, entertainment, and OTT leaders.

The company recently announced a new real-time CDN observability capability across companies’ distributed infrastructure, from edge to enterprise. This opens their solution for use across any of the major CDN providers. It also solves a critical need for anyone using a multi-CDN strategy, providing critical real-time correlation. While Hydrolix is known across the streaming market, video customers aren’t the only ones they support. The company also pulls in data tied to ad tech, cybersecurity, AIops and ITops. It is deeply integrated with AWS, offering a managed service for origin-to-edge observability for AWS CloudFront, AWS WAF, and AWS Elemental and with Akamai’s Connected Cloud, helping their customers detect and address performance and security issues proactively.

Hydrolix’s CEO Marty Kagan, co-founder of Cedexis in 2008, is a name many readers from the streaming industry will remember. Cedexis’s flagship product, Radar, monitored the quality of experience delivered by third-party CDNs to their customers, which was later acquired by Citrix in 2018. Before Cedexis, Marty spent over 8 years at Akamai. Across the company, there are some really smart people at Hydrolix working to solve big data problems for multiple industries and some of the largest Fortune 500 companies. I like what Marty and the Hydrolix team are building, the way they are going about it and the importance of their data platform that powers these critical data-intensive applications.

In Q1 of 2025, Hydrolix raised $80 million in a Series C round and has raised $140M to date. The company crossed $50 million in annual recurring revenue (ARR) in April of 2025. If you want to learn more about Hydrolix, you can listen to my podcast with Marty, recorded earlier in the year, before I was a shareholder.

Disclaimer: I have never bought, sold, or traded any shares in any public CDNs, and even in my managed portfolios, CDNs are excluded. I was previously a shareholder in MediaPublisher and Encoding.com through share grants as an advisor, not via an investment. Aside from Datazoom and Hydrolix, I do not own any other shares of any other vendor in the streaming media industry.

Netflix just made its service less user-friendly when traveling. Netflix announced it no longer supports casting shows from mobile devices to most TVs and streaming devices. Casting will now only work on older Chromecast models without remotes and TVs with Google Cast, regardless of whether you’re on an ad-supported or ad-free Netflix plan.

This is a bad decision, and it will impact users who stream Netflix in hotel rooms, where the hotel often only allows casting and doesn’t let you plug a device into the TV’s HDMI port. Even when you can access the HDMI ports on hotel TVs, the remote is often limited in functionality and won’t let you switch inputs. For this reason, I always throw Samsung, LG, and Vizio remotes in my bag. This change will also affect users who cast to a projector where no native app is available.

Netflix hasn’t said why it is dropping support for casting, but it is an odd approach for a company that says it always thinks about the service from a user standpoint. I can only guess that the motive was to make it harder to share accounts since mobile devices are a loophole around account-sharing restrictions.

Due to AI buildouts, the cost of memory and SSDs is skyrocketing, even for consumer SSDs and gaming hardware. Four months ago, I bought a SanDisk 1TB Portable SSD for $79.39. Today it costs $109.99, although Amazon currently has it on sale for $99. Thanks to buildouts for AI workloads, both DRAM memory used for RAM and NAND memory used for SSDs are in short supply. To put it in context, to support OpenAI’s massive Stargate data center initiative, the company signed deals with Samsung and SK Hynix for 900,000 DRAM wafers per month, which could amount to 40% of current global DRAM output.

According to CyberPowerPC, “global memory (RAM) prices have surged by 500%, and SSD prices have risen by 100%,” forcing them to raise pricing on PC gaming builds. Even though SSDs in AI servers aren’t the same as those in gaming PCs, NAND production for these AI server SSDs can divert from that for consumer SSDs once supply is limited. In September, Western Digital announced HDD price increases and shipping delays of up to 10 weeks, while SanDisk raised NAND pricing and Micron implemented a week-long pricing freeze.

Now, the gaming industry is watching to see if graphics cards are next to see a significant price increase, since the same production facilities that make memory for AI servers and RAM/SSD also produce graphics cards. AMD has already announced small GPU price increases, with more expected.

Netflix’s latest tech blog details how AV1 now powers approximately 30% of all Netflix viewing (on demand), following the launch of AV1 support on Android in 2020. While H.264/AVC remains the primary codec for Netflix viewing, the company expects AV1 to become the top codec very soon. Some other key takeaways from Netflix’s post:

• AV1 sessions use one-third less bandwidth than both AVC and HEVC, resulting in 45% fewer buffering interruptions
• On average, AV1 streaming sessions achieve VMAF scores that are 4.3 points higher than AVC and 0.9 points higher than HEVC sessions
• 85% of Netflix’s HDR catalog (from the perspective of view-hours) has AV1-HDR10+ coverage, and this number is expected to reach 100% in the next couple of months
• The AV1 specification incorporates a unique solution called Film Grain Synthesis (FGS), allowing Netflix to deliver a realistic cinematic film grain experience without the usual data costs
• Netflix is evaluating the use of AV1 in live streaming to deliver high-quality live experiences to large audiences without compromising video quality, and to reduce its delivery costs
• AV1 offers an opportunity to make graphic overlays highly customizable, since layered coding is supported in AV1’s main profile
• Over the past five years (2021–2025), 88% of large-screen devices, including TVs, set-top boxes, and streaming sticks, submitted for Netflix certification have supported AV1, with the vast majority offering full 4K@60fps capability

Netflix says it is excited about the forthcoming release of AV2, announced by the Alliance for Open Media in September, with an expected release at the end of this year.