Srinivas Varma

Co-founder & CEO @ProcEzy.

Passionate about all things Tech and Foreign Policy.

[Srinivas Varma @sj_varma]

My take on why you should probably just go for a windows home license if you don’t have extensive security or compliance requirements.

[Srinivas Varma @sj_varma]

@aryan34_35 Also they have to make sure that the features they deliver work best on their new hardware, own the inference stack for their models.

[Srinivas Varma @sj_varma]

Vertical integration can lead to a better cost structure for OpenAI clients.
Now that the mid to lower end models are increasingly being commoditised i think we will be seeing a price war soon in that space

[Srinivas Varma @sj_varma]

@aryan34_35 Absolutely. I have a feeling they'll not go through with the username feature.

[Srinivas Varma @sj_varma]

Im a Fan of the key feature though, i think they should just roll that out regardless of rolling out the Username feature or not. Maybe make it 6 digit alpha numeric instead of 4 digit numeric.

[Srinivas Varma @sj_varma]

@aryan34_35 Apart from lock-in's i think the real retention levers are going to be rapid product development and price.
Lock ins make customers feel a bit trapped and helpless but the other two should keep them genuinely happy

[Srinivas Varma @sj_varma]

DRAM prices have been surging and its primarily a supply chain issue.

This impacts your IT hardware budget for the next 2-3 years atleast.

[Srinivas Varma @sj_varma]

Are Frontier AI models going to be used as foreign policy bargaining chips ?

[Srinivas Varma @sj_varma]

Apple just hiked their prices. Contact your vendors and try to lock in old prices for their existing stock if possible.

[Srinivas Varma @sj_varma]

AMD's new product launch was really interesting but mass adoption might face a slight hurdle......watch for details.

[Srinivas Varma @sj_varma]

IT hardware buyers, heads up - HP raised prices 10% from April 1. Lenovo reprices any unshipped order after June 30. Check Point added 5% surcharge.
That's because, AI data centres are eating chip supply.
Indian IT teams that moved early in 2021-22 saved 18-25%, pattern repeats.

[Srinivas Varma @sj_varma]

https://t.co/UavVsjHdaW

[Srinivas Varma @sj_varma]

IT procurement is getting more expensive and complex in 2026. Key updates:

* Server & laptop prices up 15–20% (Dell, Lenovo, HPE)
* VMware renewals seeing up to 5x increases
* Microsoft M365 prices rising up to 33% in July

More in ProcEzy Weekly 👇

[Srinivas Varma @sj_varma]

Hi Folks,

I've been speaking to a few Startups recently and seems like everyone has had some kind of challenge with IT procurement at some point.

Im curious - Whats the biggest issue you've faced while trying to buy IT hardware or software for your company ?

Trying to understand pain points from people who deal with this firsthand.

[Srinivas Varma @sj_varma]

@garrytan @Replit @emergentlabs @Taskade Are people forgetting that it takes more than just coding a UI to running an app, especially something that's business critical and needs constant uptime.

Vibe code is great and all but you also need to host and maintain....cost of that vs the convenience of SaaS.

[Srinivas Varma @sj_varma]

Getting my hands on 5090 cards has been an issue of late. Anyone else facing the same issue

Also whats with this new form factor ?

[Srinivas Varma @sj_varma]

Model compression is where the real innovation is happening in AI engineering.

Smaller faster models designed to run in the edge is the precursor for affordable robotics at home.

Think having a modem sized personal assistant that stores all data in house and not in cloud, that would mitigate the privacy issues of having cloud connected AI devices as well.

[Srinivas Varma @sj_varma]

Although googles TPU's are making massive progress, this post from NVIDIA clearly talks about how they are still very much ahead, a generation lead is a big deal and can take a year or so to catch up to, even with the resources of google. Such is the nature of the Hardware game.

But interesting times ahead and competition is getting hot.

[NVIDIA Newsroom @nvidianewsroom]

We’re delighted by Google’s success — they’ve made great advances in AI and we continue to supply to Google.

NVIDIA is a generation ahead of the industry — it’s the only platform that runs every AI model and does it everywhere computing is done.

NVIDIA offers greater performance, versatility, and fungibility than ASICs, which are designed for specific AI frameworks or functions.

[Srinivas Varma @sj_varma]

Google doubling down on TPU's shows how fast AI compute is fragmenting

If tpu's keep giving better price performance than gpu's then thos people who are tired of the nvidia monopoly will have a fun time.

AI hardware is finally getting some competition