Why Spam Calls Still Reach You: India's DND Gap Explained

Your phone buzzes at 9:40 on a Tuesday. Unknown number, ten digits, local prefix. You decline it. It rings again from a different number before you have finished the sentence you were typing. You registered on Do Not Disturb, or you think you did, or you did it years ago on a phone you no longer own. And the calls keep coming anyway.

TL;DR: TRAI barred or disconnected 1.83 lakh telecom resources between April and June 2026. Your phone still rings because only 16 percent of Indian subscribers have Do Not Disturb switched on. That one setting is the first thing to check, and most people never have.

Why it matters

The enforcement side of this is not for show. In its quarterly report on unsolicited commercial communication, released on 4 August 2026 and covering April to June, the regulator counted more than a lakh telecom resources barred on a first violation and tens of thousands disconnected outright for a year. Senders blacklisted. SMS headers killed. Warning notifications sent in the hundreds of thousands. Read that report on its own and you would conclude the problem is being solved.

Then you reach the coverage line. India has roughly 1.348 billion mobile subscribers. Active DND registrations sit far below that, and everyone outside them is in the default state, which is not silence, it is consent. A number that was never registered is legally open to promotional calls from every telemarketer who did sign up properly. So the regulator can bar sender after sender while an unregistered subscriber's phone keeps ringing, because nothing about those particular calls broke a rule. This site has been circling the same gap for years, back to when the TRAI complaint app did not exist on iOS at all and a large share of users had no working way to report anything.

Here is the part that annoys me. Opt-in spam protection is a settings tax, and it falls hardest on the people least likely to go hunting for the setting: older users, first-time smartphone owners, anyone whose handset was configured by a shop assistant in ten minutes. The honest design runs the other way. Make the quiet phone the default and make marketing the thing somebody has to switch on. That is not the regime we have, and until it is, the enforcement statistics will keep improving while the lived experience sits exactly where it was.

First-strike bar

15 days

on a first violation only

Operator penalty

Rs 10 lakh

per repeat misreporting violation

Calls AI flagged

22.99 billion

in one quarter alone

Official complaint rate

1.5 per million

calls carried in the quarter

The penalty figure is worth sitting with, because it shows where the pressure actually lands. That money is charged to your operator, not to the caller, and it is charged for misreporting spam counts rather than for the spam itself. Airtel, Jio and Vi are being pushed to keep honest books. Whether that push ever reaches the man dialling you from a rented SIM in a two-room office is a separate question, and one this site has poked at before while watching how operators treat prepaid customers who complain.

"

One and a half complaints per million calls is not evidence that spam is rare. It is evidence that almost nobody bothers to report it.

Put the moving parts side by side and the shape of the gap gets obvious. Every row below is doing its job. The system still leaks.

Category	Detail	Insight
Enforcement	1,37,053 resources barred and 46,786 disconnected in the June quarter	Punishment lands on senders, not calls
Coverage	224 million registered lines against 1.12 billion that never opted out	The default setting is the problem
Complaints	1.085 million filed in the quarter, 0.553 million found actionable	Half of every report goes nowhere
Reporting	89 percent of complaints arrived through the TRAI DND app	One app carries the whole system
Numbering	140xx marks promotional calls, 1600xx marks service and transaction calls	Only honest senders wear the label
Caller ID	Truecaller barred from showing user spam reports on those two series	Your best filter lost its signal
Window	You get 7 days from the call to file a complaint	Miss it and the record disappears

The Truecaller row is the one that shifted this year. In July 2026 the company said publicly it had been stopped from displaying community spam reports on the 1400 and 1600 series, and shipped a "Frequently Blocked" badge as a workaround. Its own figures, drawn from eight months of user behaviour, showed roughly four in five of those regulated calls ignored outright and 74 million of them blocked by hand. Barring the label was the wrong call. If four out of five people already refuse to pick up, hiding that from the sixth protects no legitimate business, it just removes the only warning the reader had.

And the escalation path behind a complaint is slower than most people assume. It runs on clocks, not on outrage.

Read the timeline left to right: you file the complaint, your provider has five days to act on it, five complaints inside ten days is enough to flag the sender, and a repeat offender loses its telecom resources for a year.

Friction points

Registering on DND fixes exactly one category of nuisance, and it is worth being precise about which. It stops registered telemarketers, the ones dialling from 140 and 1600 numbers who have something to lose. It does nothing about the caller on an ordinary ten-digit SIM offering a personal loan, because he was never inside the system to be thrown out of it. Those calls are a fraud problem, handled through a completely different door. And the phone in question is the same one that pairs your smart home gear and holds your bank OTPs, which is precisely why training yourself to ignore it is a bad habit to build.

That other door is worth knowing before you need it. The Ministry of Home Affairs told the Rajya Sabha on 11 February 2026 that more than Rs 8,189 crore had been saved across over 23.61 lakh complaints filed through the national cyber fraud reporting system as of the end of 2025. Money gets frozen when a report lands fast, not when it lands eventually. The helpline is 1930 and the portal is cybercrime.gov.in, and neither of them is the DND app. Getting that wrong burns the hours that matter most.

The reporting concentration bothers me too. The official picture of spam in India is assembled almost entirely from one government app with roughly 3.3 million downloads, in a country with over a billion active connections. The revamped version launched on 13 February 2026, with TRAI chairman Anil Kumar Lahoti promising simpler screens, multilingual support and faster complaint registration. It is free, which somehow still needs saying. But a pipeline that narrow means the national spam statistics really describe what a small, motivated, mostly urban minority chose to report, which is a different thing from what the country actually receives. The same measurement blind spot shows up in how ad-driven services get judged, where the visible complaint is treated as the whole story.

• Category-wise DND is not full DND. Choose the block-everything option unless you genuinely want banking or property offers.
• Register from the number being spammed. A complaint filed from a different handset goes nowhere.
• Note the calling number and the time before you clear your call log. Both are required fields.
• A message promising money is a fraud report, not a preference report. Different route, different clock.
• Re-register after a SIM swap or a number port. Preferences do not always survive the move.

Switch it on properly

Full block, from the affected number, checked again after any SIM change.

Report inside the window

Number, date and time, filed while the call is still in your log.

Keep a labelling app

It stays your only filter on unknown ten-digit numbers.

So do the boring thing today. Open the TRAI DND app, register the number you actually use, pick the full block rather than a category list, and file the next junk call inside the week instead of muttering at it. The regulator has proved it will act on what it is told about. It cannot act on the call you never reported, from a phone it has no reason to believe wants quiet.

RAM And SSD Prices In 2026: Upgrade Now Or Wait

Your laptop takes forty seconds to open a spreadsheet. You look it up, and every answer says the same thing it has said since 2005: add more memory, swap the drive for an SSD. So you open a shopping tab, price a 32GB kit, and stop. The number is wrong. Not slightly wrong. It is roughly triple what the same kit cost the last time you checked, and nothing about your laptop has changed.

TL;DR: AI data centers have absorbed the memory supply that used to go into consumer parts, so the standard "just add RAM" fix now costs three times what it did. Do the free tuning first. Buy only if your machine is genuinely starved, and buy small.

Why the old upgrade advice stopped working

Here is the mechanism, without the supply-chain poetry. High Bandwidth Memory for AI accelerators is built on the same wafer capacity as the DDR5 in your desktop and the NAND in your SSD. When a hyperscaler orders memory by the truckload at a margin no retail buyer can match, manufacturers reallocate. Micron has already killed off its Crucial consumer memory and SSD lines to concentrate on HBM and enterprise customers. That is not a company reacting to a bad quarter. That is a company reading a decade.

The scale of it is what catches people off guard. Cameron Crandall, Kingston's Datacenter SSD Business Manager, laid the numbers out on The Full Nerd Network podcast, and the detail worth noticing is not the headline increase so much as its shape: roughly 70% of the rise since early 2025 arrived inside one 60-day stretch. That is not a trend line. That is a step. Because raw NAND flash dominates what a solid-state drive costs to build, no amount of clever engineering absorbs a jump like that on your behalf. It lands on the sticker.

SHORTAGE HORIZON

2027+

Per Samsung and SK hynix

32GB DDR5 FLOOR

$375

Cheapest retail kit

NAND COST RISE

246%

Since Q1 2025

SSD BILL OF MATERIALS

90%

Is raw NAND flash

That last figure is the one worth sitting with. When a single input dominates a product's cost that heavily, the manufacturer has no room to eat an increase on your behalf, and no incentive to try. Retailers pass it straight through. It also means storage and memory behave nothing like the rest of your machine, where a CPU generation can get faster and cheaper in the same year.

And this is where I part company with the usual advice. Rising RAM and SSD prices have made "buy before it gets worse" the standard line, which is what you would expect vendors to say and what plenty of enthusiast forums are repeating. For a small number of people that is correct. For most, it is panic dressed as strategy. If your machine is slow because forty browser tabs and a background sync client are fighting over 8GB, more memory papers over a habit. The same logic applies to the storage squeeze we covered in why Apple's 5GB iCloud tier stopped being defensible: the constraint is real, but the reflex to pay your way out of it is usually the expensive answer. Worth checking what is actually consuming your memory before you assume the hardware is the problem, because modern AI-assisted browsers are far hungrier than the ones they replaced, something that came up in our comparison of Comet Browser and Chrome.

Buy now or wait: what each choice actually costs

The decision is not really about predicting the market. It is about how much runway your current machine has and how much of a price premium you are willing to pay for certainty. Those two things point in opposite directions for most people, so it helps to see the trade-offs side by side rather than as a single verdict.

Dimension	Buy now	Wait it out
Price direction	Still rising, but slower each quarter	No projected decline before 2027
Risk if you are wrong	Overpaid on a part you barely needed	Same part costs more in six months
Free alternatives first	Usually skipped in a panic buy	Forced, and often enough on their own
Capacity discipline	Tempting to over-buy "while you are there"	You buy exactly what you ran out of
Supply availability	Consumer lines thinning as brands exit	Fewer models, more own-brand substitutes
Second-hand market	Ignored, though prices moved there too	Realistic option if you can verify wear
Whole-system pressure	Component now, laptop price later anyway	Repair beats replacement while it lasts
Best suited for	A drive down to its last few gigabytes, or a half-finished build	Anyone whose machine is slow but not actually out of room

Read the bottom row first and the rest of the table makes sense. Buying now is defensible when the shortage is blocking something concrete, like a machine that physically cannot install its next update, or a PC build sitting unfinished on a desk. It is a bad idea when the only thing pushing you is a headline. The deceleration below shows why the urgency has softened.

Quarter-over-quarter contract price increases, per TrendForce's July 2026 memory pricing survey: the climb is flattening, though it has not reversed.

Where this bites, and what to check first

The flattening comes with an asterisk that most coverage skipped. TrendForce is explicit that the slowdown is not caused by better supply. It is caused by device makers hitting the ceiling of what they can pass on to buyers who have stopped saying yes. Supply is exactly as tight as it was. Demand blinked. That distinction matters because a demand-driven pause reverses the moment inventories clear, whereas a supply-driven one tends to hold.

Which brings up the genuine grey area, and I do not think anyone can resolve it honestly. The same TrendForce numbers support two readings. One says the curve has topped out and the worst is behind us. The other says buyers are simply exhausted, and the climb resumes as soon as the current stock of pre-shortage inventory runs down. Both are consistent with the data. Anyone telling you which one is right is guessing with more confidence than the evidence supports, which, to be fair, is most of what passes for buying advice in a shortage. Before you spend anything, work through the list below.

• Check actual memory pressure, not the percentage. A machine sitting at 80% RAM use is behaving normally. Sustained swapping to disk is the real signal, and Task Manager or Activity Monitor will show it.
• Reclaim disk space before buying storage. Cache folders, old system restore points and downloaded installers routinely account for tens of gigabytes on a drive its owner believes is full.
• Audit what launches at boot. Sync clients, updaters and chat apps you never open are the most common cause of a machine that feels starved without being starved.
• Buy the capacity you need, not the capacity you want. Over-buying in a shortage compounds the damage, and the same restraint applies to any hardware purchase, which is the case we made in the Wi-Fi 7 and Wi-Fi 6E upgrade comparison.
• Verify the part fits before ordering. Thinner laptops increasingly solder memory in place, and a soldered board turns a memory upgrade into a whole-machine decision.
• Treat "certified" and "compatible" as separate claims. A stick that matches the spec sheet can still fail to post at rated speed, the same badge-versus-behaviour gap we hit with Matter-certified smart home gear.

Index card: three things the headlines left out

✔ PC makers stockpiled client SSDs through the first half of 2026, and that inventory is why suppliers have turned flexible on storage contracts while memory stays rigid.

✔ Retail USB drives and memory cards are moving slowly, because their upstream cost increases are the hardest to pass on to shoppers. Discounts turn up here first.

✔ Brands leaving the consumer market do not stop making chips. They keep making them for data centres, which is why the shelf empties faster than the fabs slow down.

So: spend an evening on the free fixes before you spend anything else. Clear the drive, kill the startup clutter, watch whether the machine actually swaps under real work. If it does, and you are genuinely out of headroom, buy the smallest part that solves the specific problem and no more. RAM and SSD prices are not going to reward patience this year, but they punish over-buying far harder than they punish waiting.

Matter Smart Home In 2026: What Works, What Still Breaks

You buy a $15.99 smart plug, scan the QR code, and your phone thinks about it for ninety seconds before giving up. Second try, same result. The box shows four logos — Apple Home, Google Home, Alexa, SmartThings — and every one of them is telling the truth. What the box doesn't say: the plug only speaks 2.4 GHz, your phone is parked on 5 GHz, and the pairing handshake quietly dies somewhere in between. That's the Matter era in one scene. Truly better than what came before. And still occasionally maddening.

TL;DR: Matter smart home support is finally broad enough to build on — thousands of certified devices, four major platforms, prices matching non-Matter gear. But pairing failures, missing advanced features, and brand holdouts persist. Buy Matter for new basics; keep native apps for anything complex.

Quick refresher for anyone who tuned out after the hype cycle. Matter is the shared smart home language run by the Connectivity Standards Alliance and backed — jointly, on purpose — by Apple, Google, Amazon, and Samsung. One certification badge replaces three "works with" stickers. A certified bulb pairs with whichever platform your household already uses, and you're no longer betting your light switches on which voice assistant wins.

The spec has grown up fast. Matter 1.3 (May 2024) added energy tracking and EV charger support; Matter 1.4 (late 2024) folded in batteries, solar systems, and heat pumps. The pace hasn't slowed since. Industry trackers such as the German trade site matter-smarthome, in its 2026 status review, point to the Matter 1.5 release expected in December 2026 — the one bringing native camera support — with more than 4,200 certified device types anticipated by then. Cameras were the loudest missing piece for four years. That gap is finally closing.

Price was the other objection, and it's gone. IKEA now sells a Matter-ready TRÅDFRI bulb for $7.99 and a PARASOLL door sensor for $9.99; TP-Link's Tapo Matter plug runs $15.99. Certified gear costs the same as the proprietary stuff sitting next to it on the shelf. Because the premium disappeared, the decision is no longer "is Matter worth extra money" — it's "is Matter mature enough to standardize on." The numbers below frame that question honestly.

Certified Devices
4,200+
expected by Matter 1.5

Entry Price
$7.99
cheapest certified bulb

Spec Cadence
~12 mo
between major releases

Multi-Admin
4-in-1
platforms, one device

Multi-admin is the sleeper feature in that grid. It means a single plug can live in Apple Home and Google Home and Alexa at the same time — not mirrored through clunky integrations, but natively controlled by each. For mixed households (iPhone parent, Android teenager, Echo in the kitchen), this quietly ends a decade of "sorry, that switch only listens to me." Nobody has to migrate. Everybody's app just works.

So should you build on Matter, or stay inside one brand's walled garden? The honest answer depends on which trade-offs you can live with, and the two paths split most sharply on features, setup pain, and long-term risk.

Dimension	Matter (cross-platform)	Single-Platform Native
Device choice	Any certified brand, mix freely	Locked to the "works with" list
First-time setup	QR scan; hit-or-miss on first pairing	Polished, brand-tuned onboarding
Advanced features	Basics guaranteed; extras often absent	Full feature set, day one
Local control	Built in via Thread; works offline	Frequently cloud-dependent
Voice assistants	All four majors, simultaneously	Usually one, sometimes two
Price premium	None as of 2026	None, but lock-in has costs later
Brand coverage	Holdouts remain (Lutron ships no Matter lighting)	Mature, complete product lines
Long-term risk	Open standard, industry-backed	One app shutdown strands hardware
Best Suited For	New setups, mixed-platform households	Power users deep in one brand

Read that table top to bottom and a pattern emerges: Matter wins on freedom and longevity, native platforms win on polish and depth. The row that surprises people is local control. Thread-based Matter devices keep working when your internet dies — a claim plenty of cloud-first proprietary gear still can't make. Here's how the standard got to this point, and why the next release matters more than the last three.

The Matter release timeline: four shipped versions in two years, then a long runway to 1.5 — the camera update that determines whether the standard covers a whole home.

Now the friction, because there's real friction. The most common failure is the one from the opening scene: commissioning. Matter devices overwhelmingly join the network over 2.4 GHz, and modern routers love to steer phones onto 5 GHz or 6 GHz bands — we covered how aggressive that band management gets in our Wi-Fi 7 vs Wi-Fi 6E comparison. When phone and device sit on different bands, pairing can fail with no useful error message. It's fixable in five minutes once you know the cause. Nobody tells you the cause.

The second problem is quieter: feature fragmentation. A certified device is only required to expose its basic functions through Matter. On/off, brightness, temperature — universal. Energy monitoring, adaptive lighting, custom animations — frequently trapped in the maker's own app. You end up with a device that answers to four platforms but only shows its full personality in one. It's the same lesson we hit with picky living-room gear like Samsung's Q-series soundbars: the badge on the box and the behavior in your house are separate facts.

And there's a grey area with no clean answer: multi-admin permissions. When one device obeys four platforms, who owns an automation conflict? If Google's schedule says lights off and Alexa's routine says lights on, the device does both, in whatever order the commands land. No spec version fully resolves household politics. Watch for these before you commit:

• Pairing fails silently — put your phone on the 2.4 GHz band (or split your SSIDs) before scanning the QR code
• "Certified" is not "full-featured" — check which functions actually surface through Matter before buying, not after
• Thread needs a border router — a HomePod mini, recent Nest Hub, or newer Echo does the job; without one, Thread devices can't reach the network
• Holdout brands break the dream — some big names still sell zero Matter gear in key categories, so audit your must-have brands first
• Firmware still flows through the maker's app — keep it installed even if you never open it for daily control

Index Card: What To Actually Remember

✔ A full starter test — bulb, sensor, plug — costs about $34 in 2026. Cheap enough to try before committing a whole house.

✔ Zero extra hubs needed if a recent smart speaker or display already lives in your home — it's probably a border router.

✔ December 2026 is the date to circle: camera support arrives in the spec, closing the last major device gap.

The blunt next step: stop buying proprietary-only devices today. Every new bulb, plug, sensor, or lock you add should carry the certification badge, because it costs nothing extra and keeps every future door open. Migrate nothing yet — your existing gear works, and replacements can wait for the camera update to prove itself. A Matter smart home in 2026 isn't finished. It's simply the only version still worth building toward.