Cost per Gigabyte

Overview

You're building a gaming PC, and you need 2TB of fast storage. The Samsung 990 Pro sits at $130—about 6.5 cents per gigabyte. Right next to it, there's a TeamGroup MP33 for $85, coming in at 4.25 cents per gigabyte. Your brain does the math. Same capacity, 35% cheaper per GB. Easy decision, right?

Wrong. Spectacularly wrong, actually.

You just optimized for the wrong metric and bought a DRAM-less QLC drive that'll slow to a crawl the moment you fill it past 50%. Your shiny new gaming rig is now bottlenecked by storage that costs you $45 less but will frustrate you for years.

Cost per gigabyte—the metric every SSD listing shoves in your face—just convinced you to make an objectively terrible purchasing decision.

This happens thousands of times daily across Amazon, Newegg, and Micro Center. Buyers optimize for $/GB because it's simple, it's everywhere, and it feels mathematically correct. But here's what the storage industry won't emphasize: cost per gigabyte became the dominant metric during the spinning rust era when all hard drives did basically the same thing at basically the same speed. It made perfect sense then. It's actively misleading now in a world where SSDs can differ in sustained performance by 10x or more at the same capacity.

I'm going to walk through exactly when $/GB optimization makes sense, when it's neutral, and when it'll waste your money on hardware that can't do what you need. We'll look at real shopping scenarios, actual performance data, and alternative frameworks that actually match what SSDs do versus just how many gigabytes they claim to hold.

How We Got Here: The Hard Drive Logic That Stuck Around

Cost per gigabyte made total sense in 2010. You walked into a store looking at spinning hard drives—everything ran at 7200 RPM, sequential speeds clustered around 100-150 MB/s, and random performance was universally terrible because, well, spinning platters with physical heads seeking across magnetic surfaces. A 1TB drive cost about $50. A 2TB drive cost about $90. The math was simple and genuinely useful. The 2TB drive gave you twice the capacity at 80% more cost—better value per gigabyte, and the drives performed basically identically beyond that.

Even early SSDs followed this pattern. First-generation SATA drives from 2011-2013 were all using similar controllers and MLC NAND. A 120GB Crucial M4 and a 120GB Samsung 840 Pro performed within the same ballpark. Picking the cheaper cost per gigabyte usually got you comparable hardware. The metric worked because the underlying technology was relatively homogeneous.

Then the NAND manufacturers discovered they could cut costs dramatically by moving from MLC to TLC to QLC, each transition roughly halving endurance and slashing sustained write performance. Controllers evolved from simple SATA chips to complex multi-core processors with DRAM buffers and increasingly sophisticated firmware. Form factors multiplied from 2.5-inch SATA to M.2 NVMe with four PCIe generations each offering dramatically different bandwidth. Cache algorithms became make-or-break features determining whether a drive maintained performance under load or collapsed into unusable stuttering.

The technology diverged wildly. The shopping metric stayed exactly the same.

Now you can find two 1TB NVMe drives both listing at 5 cents per gigabyte where one has DRAM cache, TLC NAND, and a high-end controller that maintains 3,000 MB/s sustained writes, while the other is a DRAM-less QLC drive that drops to 200 MB/s the moment its tiny SLC cache fills. Cost per gigabyte treats these as equivalent. They're not even playing the same sport.

Where Cost per Gigabyte Actually Works

Let me be clear—$/GB isn't universally worthless. There are shopping scenarios where it remains a useful primary metric, and ignoring those would be dishonest.

If you're buying mass storage for archival purposes, media libraries, or backup drives where performance barely matters and you're genuinely just paying for capacity, cost per gigabyte optimization makes perfect sense. A 4TB Samsung QVO at 5.5 cents per GB versus a 4TB Crucial MX500 at 6.2 cents per GB? For photo archival where you're reading occasionally and writing in large sequential chunks? The QVO is probably the right call—you're actually buying gigabytes, and that's genuinely what you need.

The same logic applies when comparing drives within the same performance tier. Shopping between the WD Black SN850X, Samsung 990 Pro, and Crucial T700 for a gaming build? These are all high-end TLC drives with DRAM and similar real-world performance characteristics. At that point, yeah, cost per gigabyte becomes a perfectly reasonable tie-breaker alongside warranty terms and your tolerance for each manufacturer's quirks. You're comparing apples to apples, and $/GB tells you which apple costs less.

But here's where people go wrong: they apply this same logic across performance tiers, across different NAND types, across drives designed for fundamentally different workloads. That's where the metric shifts from useful to actively destructive.

The Gaming Drive Disaster Scenario

Let's walk through the specific mistake I opened with—buying a gaming drive based primarily on cost per gigabyte. Modern games average 50-100GB installations with constant streaming of texture data, shader compilation, and asset loading. Your storage drive is a primary determinant of load times and in-game asset streaming performance.

The TeamGroup MP33 I mentioned—that 4.25 cents per gigabyte bargain? It's got approximately 12GB of SLC cache on the 1TB model. Twelve gigabytes. Once you exceed that during a large game installation or update, write speeds collapse to sub-300 MB/s according to independent testing from TechPowerUp and Tom's Hardware. Random read performance was never great to begin with thanks to the DRAM-less architecture, but under sustained load it becomes genuinely painful. Your game installations take three times longer than they should. Shader compilation stutters. Large game updates turn into hour-long ordeals instead of 20-minute downloads.

The Samsung 990 Pro at 6.5 cents per GB? That's got approximately 120GB of intelligently managed SLC cache, a full DRAM buffer, and TLC NAND that maintains 2,500+ MB/s sustained writes even when you push the drive hard. Independent benchmarks from AnandTech and TweakTown confirm sustained performance that remains consistent under load.

Real-world gaming impact: Testing from Digital Foundry and PC Gamer shows that high-end NVMe drives can reduce initial load times by 15-25% in titles like Cyberpunk 2077, Starfield, and Baldur's Gate 3 compared to budget DRAM-less alternatives. In-game asset streaming (texture pop-in, world loading during fast travel) shows measurable improvements on faster drives, particularly in open-world titles.

DirectStorage caveat: While Windows 11's DirectStorage API promises even greater performance benefits by eliminating CPU decompression bottlenecks, adoption remains limited as of late 2024. Only a handful of titles (Forspoken, Ratchet & Clank: Rift Apart) currently implement it, though more are expected in 2025. When implemented, DirectStorage can show 10-30% faster load times on high-end NVMe drives, but this isn't yet a widespread gaming benefit.

These same performance principles apply to PlayStation 5 and Xbox Series X storage upgrades—both consoles require sustained sequential read speeds above 5,500 MB/s for PS5 and consistent performance for Xbox, making budget DRAM-less drives poor choices despite their attractive $/GB.

You saved $45 optimizing for cost per gigabyte. You're now living with measurably worse performance in the exact workload you built the PC for. The math worked. The result sucks.

Quick Decision Framework

Before diving deeper, here's how to quickly filter drives for your use case:

Gaming (Primary Drive)

• Prioritize: TLC NAND, DRAM cache or good HMB, 500K+ random read IOPS, PCIe 4.0

• Avoid: QLC, DRAM-less budget drives, small cache (

• Budget Range: $80-150/TB

Content Creation (Video/Photo)

• Prioritize: High sustained write (2000+ MB/s), large cache (100GB+ on 1TB), 600+ TBW rating

• Avoid: QLC, drives with

• Budget Range: $100-200/TB

General Use / Secondary Drive

• Prioritize: TLC NAND preferred, decent cache, 3+ year warranty

• Avoid: Extremely cheap QLC, no-name brands

• Budget Range: $50-80/TB

Mass Storage / Archives

• Prioritize: Capacity, 3+ year warranty, established brand

• Avoid: Performance specs don't matter much

• Budget Range: $35-60/TB

The Performance per Dollar Framework

If cost per gigabyte misleads by ignoring what drives actually do, what's the alternative? Performance per dollar—measuring the actual capability you're buying relative to cost. This requires slightly more effort than dividing price by capacity, but it matches reality far better for non-archival workloads.

For gaming and general use, sustained random read performance matters most. A drive that delivers 600K random read IOPS at $100 gives you 6,000 IOPS per dollar. A drive delivering 400K IOPS at $80 gives you 5,000 IOPS per dollar. The cheaper drive by absolute cost is actually worse value for gaming workloads—you're getting less of what actually matters per dollar spent.

Content creators should optimize for sustained write performance instead. Video editing with 4K footage means constantly writing large files while scrubbing timelines—you need drives that maintain high sequential write speeds under sustained load. A drive with 200GB of SLC cache maintaining 3,000 MB/s sustained writes at $150 delivers 20 MB/s per dollar of sustained performance. A $100 drive with 30GB of cache that collapses to 800 MB/s gives you 8 MB/s per dollar. Three times worse value despite better cost per gigabyte.

Practical example: When editing 4K 60fps footage, a project might involve 50-100GB of footage captured in a session. Budget drives with small caches will hit cache exhaustion within minutes of timeline scrubbing, causing stuttering playback and significantly slower export times. High-end drives maintain consistent performance throughout multi-hour editing sessions.

This framework isn't perfect—you're still somewhat arbitrarily choosing which performance metric to optimize for, and most manufacturers don't publish sustained performance numbers that would let you calculate this easily. But it fundamentally reorients thinking from "how many gigabytes am I getting" to "what performance am I actually buying for this workload."

Total Cost of Ownership Changes Everything

Here's where cost optimization gets genuinely interesting—when you account for drive longevity, replacement costs, and the value of not dealing with hardware failures. A $100 TLC drive rated for 600 TBW (terabytes written) versus a $70 QLC drive rated for 200 TBW looks different when you actually use these drives.

Gaming workloads typically generate 10-20 TB of writes annually between game installations, updates, and shader cache management. Content creators can easily hit 50-100 TB annually with 4K video projects. That QLC drive reaches its endurance rating in 2-4 years depending on workload. The TLC drive lasts 6-12 years at the same usage patterns.

Important note: Rated TBW is conservative—most drives exceed their rated endurance by 2-3x in testing before failure. However, using a drive beyond its warranty coverage means you're accepting increased failure risk without manufacturer support.

Now factor in replacement hassle. Cloning drives, reinstalling operating systems, reconfiguring software, dealing with potential data loss if the drive fails suddenly rather than gracefully—there's real time cost here, and for many people, time is worth more than the $30 saved buying the cheaper drive. A drive that lasts three times longer (within warranty) effectively costs one-third as much when amortized over its useful life.

Premium drives also tend to include better warranty terms—five years versus three, with more responsive support when things go wrong. Samsung and WD generally replace failed drives faster than budget manufacturers, and the high-end models are more likely to include data recovery services. These aren't line items on a spec sheet, but they're real value that cost per gigabyte completely ignores.

You can mathematically justify spending 50% more per gigabyte if you're getting 3x the warranty coverage and significantly lower hassle over the drive's lifetime. The pure $/GB calculation misses this entirely because it treats all gigabytes as equally durable and equally likely to survive long-term.

When Marketing Exploits the Metric

Here's where cost per gigabyte becomes actively problematic—when manufacturers use it to obscure fundamental performance or reliability differences. You'll see drives marketed aggressively on capacity and price with no mention of NAND type, cache configuration, or sustained performance characteristics. "2TB for $89.99—just 4.5 cents per gigabyte!" Sounds compelling until you discover it's a QLC drive with 12GB of cache and no DRAM buffer.

Retailers participate in this by defaulting sort options to "price: low to high" or "best value" calculated purely on capacity per dollar. The algorithm actively surfaces the worst-performing drives because they optimize the metric customers have been trained to use. Amazon's "Recommended for you" based on your shopping cart? It's comparing cost per gigabyte across wildly different performance tiers and steering you toward cheaper drives that may be completely inappropriate for your needs.

Some manufacturers have started listing "effective capacity" or "usable capacity" that's lower than the actual drive size to make their $/GB look better by comparison—claiming that a 2TB drive really offers "2.2TB" because of overprovisioning or compression. This is pure marketing nonsense designed to game the metric rather than provide useful information.

The industry knows cost per gigabyte can be misleading for modern SSDs. They use it anyway because it benefits whoever makes the cheapest product regardless of performance. As long as consumers keep optimizing primarily for this metric, manufacturers have limited incentive to transparently communicate what actually differentiates their products.

Red Flags When Shopping

Watch for these warning signs that indicate you're looking at a budget drive being sold primarily on $/GB:

Marketing language to avoid:

• "High capacity at an affordable price" (without performance specs)

• "Perfect for everyday computing" (code for slow)

• "Entry-level NVMe" (DRAM-less, small cache)

• Specs listing only sequential read/write (hiding poor random performance)

• No mention of NAND type (almost always QLC)

• "Advanced caching technology" (marketing speak for small SLC cache)

When browsing Amazon or Newegg:

1. Filter search results by customer ratings AND read the 3-star reviews (most honest feedback IMO)

2. Look for technical specifications in the product listing: NAND type (TLC vs QLC), DRAM presence, cache size

3. If specs aren't listed, search the model number + "review" to find professional benchmarks

4. Compare warranty terms: 3 years is standard, 5 years indicates manufacturer confidence

5. Check the manufacturer: Samsung, WD, Crucial, SK hynix are established; unknown brands are riskier

Comparison shopping tips:

• Never sort by "price: low to high" as your primary filter

• Use specification filters for "Interface: PCIe 4.0" and look for TLC in descriptions

• Cross-reference model numbers with reviews from Tom's Hardware, AnandTech, or TechPowerUp

• Compare drives within the same capacity tier (don't compare 500GB to 2TB pricing)

Real Drive Comparison

Here's what the numbers look like when you compare actual drives across tiers:

Samsung 990 Pro (1TB) - Premium Gaming/Creation

• Price: $130 | Cost per GB: $0.065

• NAND Type: TLC | DRAM: Yes

• SLC Cache: ~120GB | Sustained Write: 2,500+ MB/s

• Random Read IOPS: 600K+ | Warranty: 5 years | TBW: 600

WD Black SN850X (1TB) - Premium Gaming

• Price: $110 | Cost per GB: $0.055

• NAND Type: TLC | DRAM: Yes

• SLC Cache: ~110GB | Sustained Write: 2,400+ MB/s

• Random Read IOPS: 550K+ | Warranty: 5 years | TBW: 600

Crucial P3 Plus (1TB) - Budget Performance

• Price: $75 | Cost per GB: $0.038

• NAND Type: QLC | DRAM: No

• SLC Cache: ~30GB | Sustained Write: 800 MB/s

• Random Read IOPS: 400K | Warranty: 5 years | TBW: 220

TeamGroup MP33 (1TB) - Budget Capacity

• Price: $65 | Cost per GB: $0.033

• NAND Type: QLC | DRAM: No

• SLC Cache: ~12GB | Sustained Write: 200 MB/s

• Random Read IOPS: 300K | Warranty: 3 years | TBW: 200

Samsung 870 QVO (1TB) - SATA Storage/Archive

• Price: $90 | Cost per GB: $0.045

• NAND Type: QLC | DRAM: Yes

• SLC Cache: ~42GB | Sustained Write: 160 MB/s (SATA limited)

• Random Read IOPS: 98K | Warranty: 3 years | TBW: 360

Sustained write speeds measured after cache exhaustion. Sources: manufacturer specifications and independent testing from TechPowerUp, Tom's Hardware, AnandTech.

What this tells you:

• The TeamGroup MP33 has the best $/GB but worst performance across every metric

• The Samsung 990 Pro costs 2x per GB but delivers 12x better sustained writes and 2x better random reads

• For gaming, the performance difference justifies the cost; for archival, it doesn't

• The SATA QVO actually makes sense for pure storage despite being slower—it's designed for that workload

What You Should Actually Compare

Instead of leading with cost per gigabyte, start with workload requirements and filter for drives that can actually handle your use case. Gaming builds need high random read IOPS and reasonable sustained write performance—you're looking at TLC drives with DRAM and substantial SLC cache. Content creation requires sustained sequential writes without cache collapse—you need drives rated for high TBW with large cache buffers. Mass storage and archival can genuinely optimize for capacity per dollar since performance barely matters.

Once you've filtered to appropriate drive types, then compare within that tier using a combination of metrics:

1. What's the sustained write performance after cache exhaustion? (Look for independent reviews, not just manufacturer specs)

2. What's the rated endurance? (TBW rating per terabyte of capacity)

3. What's the warranty term? (3 years minimum, 5 years preferred)

4. What's the real-world reliability data? (Check user reviews, forums, professional roundups)

5. How does random read performance compare under sustained load? (Critical for OS drives and gaming)

Cost absolutely matters in this evaluation—you should compare what performance you're getting per dollar spent. But you're now comparing actual capabilities relative to cost rather than just dividing price by capacity and hoping performance correlates. Sometimes it does. Often it doesn't. The only way to know is to look at what drives actually do instead of just how many gigabytes they claim.

PCIe 5.0 drives are emerging but currently offer minimal real-world benefit over high-end PCIe 4.0 for gaming and creation workloads while commanding significant price premiums.

The Bottom Line

Cost per gigabyte made sense when storage technology was homogeneous and capacity was genuinely the primary differentiator. That world ended ten years ago. Modern SSDs vary so dramatically in performance, endurance, and real-world behavior that optimizing primarily for capacity per dollar is like buying a car based solely on fuel tank size—yeah, it's technically a number you can compare, but it tells you almost nothing about whether the vehicle will actually do what you need.

Use cost per gigabyte for:

• True archival and backup storage where you're genuinely just buying capacity

• Comparing drives within the same performance tier and NAND type

• Secondary storage where performance isn't critical

For everything else —gaming, content creation, OS drives, anything involving regular writes or sustained workloads—start with performance requirements and evaluate value based on what capabilities you're actually getting per dollar spent. Factor in longevity, warranty, and replacement hassle. Consider total cost of ownership rather than just purchase price.

The math that leads you to the cheapest gigabytes often leads you to hardware that can't do the job. Sometimes the best value isn't the lowest number—it's the drive that actually delivers the performance you need without making you want to tear your hair out when it inevitably cache-dumps mid-workflow. That's worth paying for, and cost per gigabyte will never tell you which drives deliver it.

Related Reading

• SSD Buying Guide 2024: Matching Drives to Workloads (link to your comprehensive guide)

• Understanding NAND Types: TLC vs QLC Explained (link to technical explainer)

• Gaming SSD Benchmarks: Do Load Times Actually Matter? (link to gaming-focused analysis)