Why Cheap Hosting Still Uses 2012 CPUs: The PE-Backed Hosting Model Explained
Disclosure: some hosting links on this page earn me a commission if you buy. Pricing and benchmark data are verified independently. Full disclosure.
HostGator's support documentation lists AMD Opteron 6376 as the CPU in their shared hosting nodes. The Opteron 6376 was released in 2012. PassMark score: 827 out of 100,000+ processors tested. ScalaHosting's shared hosting runs AMD EPYC 9474F. PassMark score: 67,000+. Both are "budget shared hosting." One runs hardware from 2012. The other runs hardware from 2023. The performance difference is not incremental. It is a 10x gap in PHP execution speed that determines whether your WordPress site loads in 80ms or 800ms. The question is not just what hardware exists, but why the gap exists and which companies are on which side of it.
The AMD Opteron 6376: What It Is and Why It Is Still Running
The AMD Opteron 6376 is a server processor released in November 2012 as part of AMD's "Abu Dhabi" generation. It was a 16-core chip built on AMD's Bulldozer architecture, manufactured at a 32nm process. In 2012, it was a competitive mid-tier server CPU. In 2026, it is 14 years old, two complete processor architecture generations behind modern AMD, and running PHP requests at approximately one-tenth the speed of current-generation server hardware.
How do I know it is still running? HostGator publishes it in their own support documentation. Their "Server Specifications" page lists AMD Opteron 6376 as the CPU for shared hosting nodes. This is not a leak or an investigative finding. The company that owns and operates the servers considers this a normal, acceptable state of affairs to document publicly in 2026.
The Opteron predates AMD's Zen architecture by five years. It predates NVMe SSD adoption, PCIe 4.0, DDR4 RAM, and most of the WordPress core updates that increased PHP resource requirements. The WordPress ecosystem of 2026 is running on hardware designed for the web of 2012.
Why Budget Hosts Don't Upgrade Hardware
The question is not technical. Replacing Opteron hardware with current-generation AMD EPYC is a solved engineering problem. Data centers replace server hardware routinely. The question is economic: who pays for it and who benefits.
For a private equity-owned hosting company, the decision calculus is specific. Server hardware replacement is a capital expenditure that reduces reported EBITDA in the year it is spent. The investor thesis for PE-backed hosting is not "grow the business by investing in infrastructure." It is "acquire a subscriber base, reduce costs, increase margins, and exit." Hardware upgrades work against that thesis in the short term, even when they would reduce churn and improve long-term subscriber retention.
For independently owned hosting companies, the decision calculus is different. The founder making the hardware investment decision is also the person whose long-term reputation and business value depends on maintaining hosting quality. There is no 5-year exit timeline pressuring short-term EBITDA. ScalaHosting is running EPYC 9474F on shared hosting because the owner decided that hardware investment was the right long-term call for the business. That decision is only possible when the owner's time horizon is indefinite rather than bounded by a PE fund's investment window.
The Full Hardware Gap: PassMark #827 vs PassMark #31
The PassMark database ranks every major CPU by composite multi-thread score. HostGator's Opteron 6376 ranks at #827. ScalaHosting's EPYC 9474F ranks at #31. Both are sold as "budget shared hosting." The hardware inside is not comparable.
| CPU | Release Year | PassMark Score | Found In | PHP Execution (est.) | Verdict |
|---|---|---|---|---|---|
| AMD Opteron 6376 | 2012 | 827 | Budget shared (HostGator documented) | ~85ms per uncached PHP request | Legacy. Avoid for any WordPress site beyond static blogging. |
| Intel Xeon E5-2650 v4 | 2016 | ~9,500 (composite) | Mid-tier shared (Bluehost documented) | ~60ms per uncached PHP request | Outdated but less severe than Opteron. Still 4-5x slower than EPYC. |
| Intel Xeon Gold 6154 | 2017 | ~22,000 | Budget VPS tier | ~25ms per uncached PHP request | Adequate for WP blogs. Below threshold for WooCommerce. |
| AMD EPYC 7302 | 2019 | ~25,000 | Mid-tier cloud VPS | ~18ms per uncached PHP request | Solid for small to mid WordPress sites. |
| AMD EPYC 7713 | 2021 | ~73,000 | DigitalOcean, Vultr | ~12ms per uncached PHP request | Strong. Current recommended minimum for performance-sensitive sites. |
| AMD EPYC 9474F | 2023 | 67,000+ (composite) | ScalaHosting shared (#31 PassMark) | ~8ms per uncached PHP request | Current flagship. What low-density shared hosting can run in 2026. |
The gap between HostGator's Opteron (~85ms PHP execution) and ScalaHosting's EPYC (~8ms PHP execution) is not the difference between "slow" and "fast" hosting. It is the difference between a server that can process a WordPress request in the time it takes to blink and a server that takes the equivalent of watching a progress bar for a single page load. Multiplied across every request your site serves, every day, the gap accumulates into measurable differences in search rankings, bounce rates, and conversion rates.
What Old CPUs Do to Your WordPress Site
The numbers above are server-side PHP execution estimates. Here is what those numbers mean in concrete WordPress site outcomes.
| Metric | Opteron 6376 (2012) | EPYC 9474F (2023) | Practical Impact |
|---|---|---|---|
| TTFB on uncached page | 300-600ms | 80-120ms | 3-7x slower |
| PHP execution per request | ~85ms | ~8ms | 10x slower |
| WooCommerce checkout | Often 800ms+ | 120-200ms | Frequently triggers cart abandonment |
| Peak concurrent users before queuing | ~8-12 simultaneous requests | 40-80+ | Lower queue tolerance under traffic spikes |
| OPcache benefit (relative) | High, but ceiling is still low | High, ceiling is much higher | OPcache helps both, but modern CPU gains more |
| WordPress admin speed | Noticeably sluggish at 20+ plugins | Responsive at same plugin count | Daily workflow impact at high plugin count |
Google's Core Web Vitals include TTFB as a component that feeds into the Largest Contentful Paint metric. A TTFB of 600ms (typical for uncached Opteron-era shared hosting) adds 600ms to every visitor's first meaningful content render. Google's "good" threshold for LCP is under 2.5 seconds. Server response time alone can consume 24% of that budget before any network or render work begins.
The correlation between TTFB and search rankings is not perfectly linear, but sites consistently serving 600ms+ TTFB face algorithmic headwinds that sites serving 80-100ms do not. The hosting hardware is a ranking factor, mediated through Core Web Vitals signals.
WooCommerce checkout pages are intentionally excluded from full-page caching because they handle personalized cart data. Every checkout request hits PHP directly. On Opteron-era hardware, WooCommerce checkout commonly delivers 800ms+ TTFB under even modest load. Cart abandonment research consistently shows abandonment rates increasing with every 100ms of checkout load time above a 300ms baseline.
If you are running WooCommerce on HostGator's shared hosting, the hardware is the direct cause of lost conversions. Not the checkout flow, not the payment gateway, not the theme. The CPU underneath the shared server that is too slow to process a checkout request before a customer's patience expires.
PHP-FPM processes requests sequentially on each worker. When requests take 85ms instead of 8ms to execute, fewer can complete per second before queuing begins. A shared hosting node on Opteron hardware begins queuing incoming requests at lower traffic volumes than the same node on EPYC hardware. A traffic spike that modern hardware handles cleanly will produce visible 502 or 504 errors on old hardware because the request queue backs up before the CPU can clear it.
There is one important caveat: full-page caching partially offsets the CPU problem. If your site has a high cache hit rate (static content, LiteSpeed cache, Cloudflare serving pre-built HTML), cached requests bypass PHP entirely and the CPU is largely irrelevant. The CPU becomes critical again for: logged-in users, WooCommerce cart and checkout, WordPress admin, and any cache miss.
How to Verify Your Host's Actual Hardware
You do not have to take your host's word about their hardware. You can verify it directly.
Take the CPU model name and search for it at cpubenchmark.net. Both the composite and single-thread scores are listed. The single-thread score predicts WordPress PHP speed; the composite score predicts parallel capacity. For shared hosting, note that you are sharing the CPU across many accounts, so the composite score does not fully translate to your available capacity.
- Create a file named
cpucheck.phpin your web root with only this content:<?php phpinfo(); ?> - Load that file in your browser via your domain
- Look for the "System" row near the top of the phpinfo output — it typically shows the server operating system and CPU model
- Copy the processor name and look it up on cpubenchmark.net
- Delete
cpucheck.phpimmediately after. phpinfo() exposes server configuration details you do not want publicly accessible.
Some hosts publish their hardware specifications directly. HostGator lists AMD Opteron 6376 in their support documentation. ScalaHosting's team has publicly documented their EPYC 9474F infrastructure including DDR5 RAM and PCIe 5.0 NVMe specs. Kinsta publishes their use of Google Cloud C2 instances. If a host cannot or will not tell you which CPU generation their shared hosting runs on, treat that as a signal.
Hosts with modern hardware tend to advertise it. Hosts with aging hardware tend not to. The absence of CPU documentation on a budget host's specification page is not neutral.
What Modern Shared Hosting Hardware Looks Like in 2026
The choice is not binary between "old hardware cheap" and "new hardware expensive." The hardware gap exists between providers at comparable price points. ScalaHosting's EPYC 9474F shared hosting starts at comparable pricing to HostGator's Opteron-era shared plans. The difference is not cost — it is ownership model and investment philosophy.
- CPU: AMD EPYC 7xxx or 9xxx series (Zen 3 / Zen 4), or Intel Xeon Scalable 3rd/4th gen (Ice Lake / Sapphire Rapids)
- RAM: DDR4 (minimum) or DDR5 (current gen). DDR5 at 4800MHz+ indicates recent server deployment.
- Storage: NVMe SSD (PCIe 3.0 at minimum; PCIe 4.0/5.0 for current-gen hosts)
- PassMark composite: Above 30,000 for the underlying CPU
- Single-thread score: Above 2,000 (above 2,500 is excellent for WordPress)
- CPU: AMD Opteron (any generation), Intel Xeon E5 series, any CPU with "v2" or "v3" suffix
- RAM: DDR3 or early DDR4 (pre-2017 deployments)
- Storage: SATA SSD or any spinning disk
- PassMark composite: Below 10,000 for the underlying CPU
- Host behavior: Refuses to disclose CPU model; hardware specifications page is absent or vague
Migration from legacy hardware to a modern host is a one-time cost that pays forward indefinitely. ChemiCloud offers free migration and guarantees 3 CPU cores on Turbo plans. ScalaHosting offers free migration with their managed WordPress plans. The performance gain is immediate and measurable in TTFB and Core Web Vitals on day one of the new host.
For anyone on HostGator, iPage, or Bluehost shared hosting who has exhausted configuration optimizations (OPcache enabled, PHP 8.2, Redis caching) and is still seeing TTFB above 400ms on uncached pages, the hardware is the remaining explanation. No plugin, no theme change, and no caching configuration changes the underlying CPU generation. Only migrating to a host with modern hardware does.
Old Hosting Hardware FAQ
What CPU does HostGator use for shared hosting?
HostGator's own support documentation lists AMD Opteron 6376 as the CPU in their shared hosting nodes. The Opteron 6376 was released in 2012. Its PassMark composite score is 827. For reference, AMD's EPYC 9474F (used by ScalaHosting in 2026) scores 67,000+ composite and handles individual PHP requests approximately 10 times faster. HostGator has not publicly documented any shared hosting hardware upgrade as of mid-2026.
How do I check what CPU my hosting plan uses?
From a VPS with SSH access, run: cat /proc/cpuinfo | grep 'model name' | head -1. This outputs the exact CPU model. Then look it up at cpubenchmark.net for both the composite and single-thread PassMark scores. From shared hosting without SSH: create a temporary file containing only and load it in a browser. Look for the 'System' row, which typically shows the server's CPU. Delete the phpinfo file immediately after viewing it. Some cPanel installations also show server hardware under 'Server Information' in the general information section.
What is a PassMark score and why does it matter for hosting?
PassMark is a standardized CPU benchmarking score from PassMark Software, with results published at cpubenchmark.net for every major CPU. The composite score measures total CPU throughput across all cores simultaneously. The single-thread score measures how fast one sequential task executes — which is the metric that predicts PHP execution speed for WordPress, since PHP-FPM handles each request on one thread. For shared hosting comparisons, the single-thread score is more predictive of real WordPress performance than the composite score.
Why would a hosting company still run 2012 hardware in 2026?
Server hardware is a capital expenditure. For private equity-owned hosting companies like Newfold Digital (which owns HostGator), hardware upgrades reduce near-term EBITDA, the metric that determines the value of the business at exit. The existing hardware still technically functions. Subscriber churn from poor performance can be attributed to other causes, obscuring the hardware correlation. Until customer acquisition cost exceeds the revenue from existing slow-performing subscribers, there is no financial pressure to upgrade. The hardware stagnates because the business model does not reward investment.
Does faster CPU always mean better hosting?
No. CPU speed is the fourth item in the WordPress performance optimization stack. OPcache (free, eliminates PHP compilation overhead), PHP version upgrade to 8.2+ (free, 15-20% speed gain), and Redis object caching (eliminates database wait time) all have larger practical impact than a CPU upgrade when they are not already in place. Fix the free optimizations first. When OPcache, PHP version, and Redis are all properly configured, single-thread CPU speed becomes the primary remaining variable for TTFB improvement.
What CPU does ScalaHosting use for shared hosting?
ScalaHosting runs AMD EPYC 9474F CPUs on their shared hosting nodes. The EPYC 9474F is a 2023-generation server processor that ranks #31 in PassMark's composite CPU benchmark database. ScalaHosting's CTO has confirmed they use DDR5 RAM at 4800MHz and PCIe 5.0 NVMe SSDs alongside the EPYC infrastructure. Running current-generation hardware on shared hosting requires founder ownership with long-term investment priorities — the opposite of what PE-backed consolidation produces.
Is it worth migrating away from a host with old hardware?
Yes, if your site generates revenue or traffic where load time matters. A 5-7x PHP execution speed improvement between Opteron-era and EPYC hardware is directly measurable in TTFB, Core Web Vitals LCP scores, and WooCommerce checkout completion rates. The migration cost, typically 2-4 hours for a standard WordPress site using a migration plugin, is a one-time expense. The performance penalty from staying on outdated hardware compounds every month, affecting search rankings, bounce rates, and conversion rates continuously.
Does Bluehost use the same old CPUs as HostGator?
Bluehost and HostGator are both owned by Newfold Digital and share infrastructure. Independent lscpu tests on Bluehost shared hosting nodes have shown Intel Xeon E5-2650 v4 CPUs (released 2016). This is newer than HostGator's Opteron 6376 (2012) but still 5-7 years old as of 2026 and significantly slower than modern EPYC infrastructure. The pattern is consistent: both brands are running hardware from the era when Newfold's PE investors acquired them, without meaningful upgrade cycles.