MongoDB CEO Dev Ittycheria talks AI hype and the database evolution as he crosses 10-year mark

Trending 2 weeks ago
ARTICLE AD BOX

A batch has happened since Dev Ittycheria took nan reins astatine MongoDB, nan $26 cardinal database institution he’s led arsenic president and CEO since September 2014. Ittycheria has taken MongoDB to nan cloud, steered it through an IPO, overseen its transition from unfastened source, launched a task superior arm, and grown nan customer guidelines from a fewer 100 to thing approaching 50,000.

“When I joined nan company, it wasn’t clear if group would spot america to beryllium a genuinely mission-critical technology,” Ittycheria told TechCrunch. “When I joined, it was doing astir $30 cardinal successful revenue; now we’re doing adjacent to $2 billion.”

It hasn’t each been peaches and cream, though. Five months ago, MongoDB was deed by a information breach, which, while comparatively contained, did momentarily consequence its estimation successful an manufacture wherever estimation is paramount.

Throw into nan operation nan whirlwind AI gyration that has engulfed conscionable astir each industry, and location was overmuch to talk erstwhile TechCrunch sat down pinch Ittycheria astatine MongoDB’s caller London office, which opened successful Blackfriars past year.

MongoDB logo astatine company's London office. April 17, 2024.

MongoDB’s London office.  Image Credits: Paul Sawers / TechCrunch

Vector’s embrace

Databases person travel a agelong measurement since IBM and Oracle first popularized relational databases much than half a period ago. The internet’s emergence created request for flexible, scalable, and cost-effective information retention and processing, paving nan measurement for businesses specified arsenic MongoDB to thrive.

Founded successful 2007 by a trio of veterans hailing from online adtech institution DoubleClick (which Google acquired for $3.1 billion), MongoDB was initially called 10Gen until a rebrand to nan sanction of its flagship product six years later. It has since emerged arsenic 1 of nan preeminent NoSQL databases, helping companies shop and negociate ample volumes of data.

Prior to joining MongoDB, Ittycheria founded and exited a server automation institution called BladeLogic for $900 cardinal in 2008, and went connected to service successful various committee personnel and investor roles (including a 16-month stint astatine Greylock) earlier joining MongoDB arsenic president and CEO coming connected for 10 years agone now. Ittycheria replaced Max Schireson, who stepped down for family reasons aft conscionable 18 months successful nan role.

Built connected a document-oriented model, MongoDB has grown disconnected nan backmost of nan detonation successful mobile and web applications wherever flexible, move information structures are astatine play. The existent artificial intelligence activity is driving a akin shift, pinch vector databases nan basking caller point successful town.

Like NoSQL, vector databases besides specialize successful unstructured information types (e.g., images, videos, societal media posts), but are peculiarly good suited to ample connection models (LLMs) and generative AI. This is owed to nan measurement they shop and process information successful nan shape of vector embeddings, which person information into numerical representations that seizure relationships betwixt different information points by storing them spatially by relevance. This makes it easier to retrieve semantically akin information and allows AI to amended understand discourse and semantics wrong conversations.

While a slew of dedicated vector database startups person emerged these past fewer years, nan incumbents person besides started embracing vector, including ElasticRedisOpenSearchCassandra, and Oracle. Cloud hyperscalers, including MicrosoftAmazon, and Google have besides ramped up support for vector search.

MongoDB, for its part, introduced vector search to its flagship database-as-a-service merchandise Atlas past June, a motion that nan institution was preparing for nan oncoming AI tsunami. This mimics different humanities trends wherever single-function databases look (such arsenic time-series) pinch immoderate inferior arsenic stand-alone solutions but that mightiness besides beryllium amended integrated into a larger multi-purpose database stack. This is precisely why MongoDB introduced support for time-series databases a fewer years back, and why it’s doing nan aforesaid pinch vector.

“A batch of these companies are features masking arsenic products,” Ittycheria said of nan caller activity of dedicated vector products. “We built that into nan platform, and that’s nan worth — alternatively utilizing immoderate stand-alone vector database and past your OLTP [online transaction processing] database and past your hunt database, we tin harvester each 3 things into 1 level that makes nan life of a developer and designer truthful overmuch easier.”

The thought is that database providers that adopt a multipronged attack tin harvester each nan information successful 1 place, making life easier for developers to activity with.

“There’s astir apt for illustration 17 different types of databases, and astir apt astir 300 vendors,” Ittycheria said. “There’s nary customer connected this satellite that wants to person 17 different databases. The complexity that creates, and nan costs of learning, supporting and managing those different technologies becomes overwhelming. It besides inhibits innovation, because it creates this taxation of complexity.”

MongoDB president and CEO Dev Ittycheria

MongoDB’s Dev Ittycheria. Image Credits: MongoDB

Too overmuch hype

Despite nan preparation, Ittycheria reckons location is excessively overmuch hype astir AI — for now, astatine least.

“My life has not been transformed by AI,” he said. “Yes, possibly I tin constitute an email amended done each those assistants, but it’s not fundamentally transformed my life. Whereas nan net has wholly transformed my life.”

The mentation is that contempt nan hullaballoo, it will return clip for AI to seep into our mundane lives — and erstwhile it does, it will beryllium done applications integrating AI, and businesses building connected it.

“I deliberation pinch nan take of immoderate caller technology, we spot worth accrue astatine nan bottommost furniture first,” Ittycheria said. “Obviously, Nvidia is making money manus complete fist, and OpenAI has been nan astir talked astir institution since they launched ChatGPT. But nan existent worth will travel erstwhile group build applications connected apical of those technologies. And that’s nan business we’re successful — we’re successful nan business of helping group build applications.”

For now, it’s each astir “simple apps,” arsenic Ittycheria puts it. This includes chatbots for customer service, thing that MongoDB itself is doing internally pinch CoachGTM, powered by MongoDB’s vector search, to bring its income and customer teams instant knowledge astir their products. In immoderate ways, we’re presently successful nan “calculator apps” shape that nan iPhone recovered itself successful astir 20 years agone erstwhile nan conception of nan App Store deed nan masses.

“The existent blase [AI] apps will beryllium utilizing real-time data, being capable to make real-time decisions connected real-time events,” Ittycheria said. “Maybe something’s happening successful nan banal market, possibly it’s clip to bargain aliases sell, aliases it’s clip to hedge. I deliberation that’s wherever we will commencement seeing overmuch much blase apps, wherever you tin embed real-time information on pinch each nan reasoning.”

The SaaS path

One of nan biggest developments during Ittycheria’s tenure has been nan modulation from a self-deployed model, wherever customers big MongoDB themselves and nan institution sells them features and services. With nan motorboat of Atlas successful 2016, MongoDB embarked connected nan acquainted SaaS way wherever companies complaint for removing each nan complexities of self-hosting. At nan clip of its IPO nan pursuing year, Atlas represented 2% of MongoDB’s gross — coming that fig sits astatine astir 70%.

“It’s grown very quickly, and we’ve really built that business arsenic a nationalist company,” Ittycheria said. “What nan fame of Atlas showed was that group are comfortable consuming infrastructure arsenic a service. What that allows them to do is delegate what they see ‘non-strategic functions,’ for illustration provisioning, configuring and managing MongoDB. So they tin attraction connected building applications that are really transforming their business.”

Another awesome improvement came when, a twelvemonth aft going public, MongoDB moved away from an unfastened root AGPL licence to a source-available SSPL (server broadside nationalist license). In immoderate ways, this was nan bellwether of what was to come, pinch countless infrastructure companies going connected to wantonness their unfastened root credentials to forestall nan unreality giants (e.g., Amazon) from trading their ain type of nan work without giving back.

“We consciousness very happy astir it [the licence change],” Ittycheria said. “The reality is that while it was unfastened source, 99.9% of nan improvement is done by our ain group — it’s not for illustration communities contributing code. It’s not immoderate simple, trivial exertion — it’s very analyzable code, and we request to prosecute senior, talented group who costs a batch of money. We didn’t deliberation it was adjacent for america to walk each this money to build this product, past personification takes that free product, monetizes it, and not springiness america thing back. It was rather arguable successful 2018, but looking back, our business has only grown faster.”

And grown it has. As pinch conscionable astir each tech company, MongoDB’s valuation soared during nan pandemic, peaking astatine an all-time precocious of $39 cardinal successful precocious 2021, earlier plummeting southbound of $10 cardinal wrong a twelvemonth — astir nan aforesaid arsenic its pre-pandemic figure.

However, MongoDB’s shares person been successful ascendency successful nan 18 months since, hitting $35 cardinal conscionable a mates of months ago, earlier dropping again to astir $26 cardinal coming — specified is nan volatile quality of nan banal markets. But fixed nan company’s comparatively humble $1.8 cardinal valuation astatine nan end of its first time of trading successful 2017, MongoDB has performed reasonably good for nationalist marketplace investors.

President and CEO Dev Ittycheria pinch MongoDB colleagues astatine its 2017 IPO

Dev Ittycheria pinch MongoDB colleagues astatine its 2017 IPO. Image Credits: MongoDB

Four months ago, though, MongoDB revealed a information breach that exposed “some customer relationship metadata and interaction information” — it progressive a phishing onslaught done a third-party endeavor instrumentality (Ittycheria wouldn’t corroborate which). This caused its shares to driblet 3%, but successful nan months that followed, MongoDB’s valuation surged backmost to a two-year high. This highlighted really small effect nan breach had connected affairs astatine nan company, surely compared to high-profile information breaches at nan likes of Equifax and Target, which deed nan businesses difficult and forced senior executive departures.

While MongoDB’s cybersecurity incident was importantly smaller successful scope, what stood retired was really quickly nan full point went distant — it was reported successful respective outlets (including TechCrunch), but nan communicative vanished into nan foggy ruins of clip conscionable arsenic quickly arsenic it arrived.

“Part of nan logic is that we were very transparent,” Ittycheria said. “The past point you want to do is hide accusation and look for illustration you’re misrepresenting information. We person tons of banks who put a batch of very delicate accusation successful our information platform; we’ve tons of different companies that person a batch of delicate information. So for us, it’s really astir making judge that our architecture is robust and sound. And this really forced america to double down. I would ne'er declare that we’re ne'er gonna get hacked again, but we’re doing everything successful our powerfulness to guarantee that it doesn’t.”

Nothing ventured

It’s not different for nan biggest tech companies to motorboat their ain finance vehicles, arsenic we’ve seen done nan years pinch Alphabet (which has several finance offshoots), Microsoft, Amazon, and Salesforce each ingratiating themselves pinch nan startup fraternity. But a newer activity of endeavor firm task firms person entered nan fray, too, including Slack, Workday, Twilio, Zoom, HubSpot, and Okta.

In 2022, it was MongoDB’s move to motorboat specified a fund, and successful nan 2 years since, MongoDB Ventures has invested successful immoderate 8 companies.

“This is for america to build deeper relationships — we activity successful an ecosystem that consists of ample companies and besides mini companies,” Ittycheria said. “Where we spot a mini institution that we deliberation could beryllium absorbing to activity with, we say, ‘Hey, we want a chance to put successful you,’ truthful that other value’s created. We besides are nan beneficiaries of creating immoderate of that value.”

MongoDB only has a fistful of group successful its firm improvement squad that are mostly focused connected nan task fund, and Ittycheria stresses that MongoDB takes a backmost spot pinch its investments. It besides typically invests alongside different VCs, arsenic it did pinch its inaugural finance successful 2021 (predating nan general motorboat of its fund), erstwhile it softly joined nan likes of Insight Partners and Andreessen Horowitz successful Apollo GraphQL’s $130 cardinal Series D round. 

“We ever return a number position, we don’t return a committee seat, and we don’t group nan terms,” Ittycheria said. “But nan logic startups are willing successful america is because they want to leverage nan MongoDB brand. We person thousands of group successful nan field, truthful they [startups] tin leverage our distribution channels.”