$GOOG recently had lots of negative sentiment, but with Gemini pro rollout, it has asserted dominance in AI space, especially huge gain in cloud growth, and core business is stronger than ever. There was some concerns but $GOOG has done well in navigating the AI space.

If top 10 remained same at the start of 6 year period then it’s a fair comparison else it’s picking winners in hindsight. For instance, according to ChatGPT, Nvidia & $AVGO were not in top 10 in 2019 & they are now, growing 5000% & 1000% respectively, contributing to the outrun.

Meta’s Ads & Microsoft’s enterprise and cloud moats are getting even stronger & keep on delivering. Considering Meta lost about $4.5B in reality labs & still grew profits by 36% & if we exclude RL losses, the net income is $24B almost same as Microsoft’s $25.8B, quite phenomenal.

There are many viewpoints that I have changed about in past 2 decades. The main reason to change viewpoints was to monitor the actions from what is being said. As actions are more important than saying & are verifiable. Without it, it’s easy to keep a biased viewpoints.

I would love to learn about the research and thoughts that went for the leadership to green light in building such a monumentally failed product. It was a solution in search of a problem.

So now India is shipping about 44% and Vietnam 30% of iPhones to US and meanwhile China’s shipments dropped to 25%. I didn’t think it was so easy to replace Chinese manufacturing capabilities moat which doesn’t seem to be true in this particular case.

AI is non deterministic; it provides knowledge & actions, but they are non deterministic & don’t face consequences of their actions. To validate AI driven code; one needs strong fundamentals, AI is an augmenter in software development & not a replacement for software engineer.

True! Apple doesn’t have a mechanism to monitor iOS app’s bad behaviour after an app is approved. It has been a privacy charade to differentiate themselves for instance in China Apple keeps data in state owned servers which is different to other regions.

Gradually, with social media & shorter form of content, the attention span is getting even shorter & socialising outside for kids doesn’t seem that rewarding than sitting on couching & consuming infinite content. I only see increased mental health issues in kids going forward.

It seems there is quite a low bar nowadays to do predictions on any topic, nobody is bothered about credibility’s. It’s not much different than horoscope I guess.

Rather investing $100B in metaverse, they could have very well created a cloud storage vertical similar to GCP and could have easily grown it to at least $20B revenue by now compared to $50B what GCP is generating annually.

Still the user needs to validate the output of AI, & in novel prompts it fails miserably, so the onus lies on the user to guide the AI with better prompts to rethink & review its context & change its output to map to user’s context. It’s not easy, & user’s own knowledge is key.

AI has most of the open world data but the problem is: it takes user prompt & matches that in realtime to its vast data & based on certain mathematical permutations & statistics & other magic, it outputs some content. If you grill it with another prompt then it backtracks.

Google has executed well in AI through their Gemini pro model & GCP growth. OpenAI had caught its search business by surpruse, now I think it is well positioned to take advantage both from AI overview, gemini pro subscription & AI offerings in GCP. Youtube, as usual, going strong.

IQ can be easily tested & measured; hence society can measure anybody's IQ & put value. Similar to IQ; individual's money net worth can be measured, hence more value is given. & like EQ, other forms of wealth can't be easily measured hence society doesn't know how to value them.

The historical shift from manual bookkeeping to excel, it changed nature of work, creating more complex & higher value work needing exponentially higher number of people. Similarly, AI would create new kind of possibilities requiring more people to work on them.

$NFLX main differentiation is creating content for local regions by collaborating directly with local talent; & especially talents that are less known to keep production cost low. Many a times local production even become worldwide hit like Korean Squid Games!

Already the average quality of the content on Netflix is low to average, it is more for masses than art. If a content is watched by at least 10% subscribers then it has done well. Due to AI the cost to predict a content will be much lower resulting even lower quality content.

It’s always better to keep things simple as much as possible as it helps at all levels. Though, complexity is a necessary evil when scale, performance & availability is needed, for a company having millions of concurrent users, one server won’t scale.

Ads is easiest form of revenue generation for apps with significant number of users, as users don’t pay & in return see ads. $NFLX recently found a new revenue growth via Ads & WhatsApp started showing Ads after trying for a decade & failing to find a comparable revenue stream.