It’s impossible not to be impressed by the stunning success of Snowflake, and as the CEO and founder of a company in a similar space, I have followed their progress with both interest and alarm! While it’s hard for me to comment on Snowflake without sounding like sour grapes, I caution any IT buyer not to make decisions based on the stock price. Warren Buffet’s investor credentials are without question, but he’s not a person I would turn to for advice on corporate analytics. Since the Snowflake stock price crash six months ago, I sense the company has embarked on a plan to appease Wall Street rather than customers.
Over to us. We started Yellowbrick Data back in 2014 and actually got the company funded in 2015. Nobody knew that three years earlier that other company, Snowflake Computing, was the “other data warehousing” startup and it had done an incredibly good job being stealthy, incubated internally within a VC fund.
I’ve mentioned before how building a successful database is really hard. Computer scientists of my era dreamed of building things like file systems and databases. I’ve still got old colleagues who to this day can’t give up building the damn things 20+ years on: They’re an addictive drug for the OCD-obsessed engineers who relentlessly pontificate about the perfect way of laying things out. Rather like organising the dishwasher and then showing off the result to everyone – “look how much I’ve packed in there, George – the plates are aligned to perfection with barely enough space in between them to fit your pinkie, and the curves on the shafts of the spoons look like a school of synchronised mermaids.” Then, seeing a small gap and realising if you re-do the whole thing, swapping the plates and the bowls and increasing the angle of lean, you can squeeze in one extra piece of crockery without it hopefully not falling over like dominoes. No wonder Linux has almost 100 file systems in it. So addictive. I digress.
Everyone’s always comparing your business to the “other start-ups” they can pattern match against. The ones they draw from are those that are most famous, those in the same industry or those that were founded around the same time. Perhaps they are the fashionable companies riding the hype cycle of buzz or a current darling of the tech world. Perhaps it’s a business that investors missed out on because they thought the valuation was too high, not realising just how much damage liquidity injected into the system was going to boost valuations to irrational levels.
In our case, that’s the Big Dog, Snowflake. An amazing company. Amazing story. The number one software IPO in history. Yep, read that again. THE. BIGGEST. SOFTWARE. IPO. IN. HISTORY. is our comparable. With a peak valuation of $120 billion dollars. If we stacked Big Dog’s market cap out of $100 bills, it would be 8,148 miles high. If we put them alongside each other, Big Dog would be occupying all Los Angeles’ land area.
There are recent signs that Big Dog is starting to lose its bark, and there’s some howling. We’ve heard from customers that their Snowpark embedded programming offering is poor, being hard to debug and seriously inferior to Databricks’ capabilities for actual processing. News articles are mentioning insiders and senior leadership selling even at a time of extreme weakness for the stock. Worried about the stock price, the Press Room has been issuing streams of press releases about crazy stuff: Connecting on-premises SANs to the cloud for “hybrid analytics” which is sheer madness, along with allegedly new vertical offerings that actually, upon careful reading, don’t really say much new. These aren’t the big issues though. The business model is the big issue.
Don’t get me wrong, a lot of technology growth businesses are having a harder time this year than in the last couple, including yours truly’s as well. However the Really Big Difference is – Snowflake’s answer to doing more is always to spend more. Want ETL running without slowing users to a crawl? Spend more to spin up more. Want to run additional workloads to add value to your business in tough times? Spin up more to spend more. There’s tons of hidden costs on top of that.
At Yellowbrick, we’ve got a business model that lets new workloads and ETL run without additional costs; things can be prioritised so that within a certain set of resources, the stuff that matters always gets done regardless. We support an order of magnitude more concurrency per cluster, full elasticity, and we let customers pay their own cloud costs so that we’re not marking stuff up and selling it back to them.
“The more we use Snowflake, the more money we spend. The more we use Yellowbrick, the more money we save” – VP of Data at one of our large customers
Big Dog should be scared with the tail down between its hind legs when its three largest competitors – Amazon, Google and Microsoft – control its COGS (cost of goods sold) and its margins. I’ve been there before, at SSD company Fusion-IO: We had the best SSD IP in the industry and the largest business in the new segment, but the raw ingredients, the flash chips, were sold by the fabs (Samsung, Intel, Micron, Sandisk and Toshiba) who were our biggest competitors. Snowflake’s biggest competitors are the vendors that sell its raw ingredients – the compute and storage that it marks up and sells back. At Yellowbrick, we don’t have this problem and the answer to everything isn’t to spend more: So we are well placed to thrive in the era of higher interest rates, more difficult borrowing, slowing business sectors and cloud cost controls.
We can’t beat Snowflake at its game, and we can’t beat their IPO, but we can definitely do well playing our own game our way for customers in financial services, insurance, telecommunications and those that want to save money at scale; the very customers that view us as the best data warehouse in the world.